Food Emulsions

Library of Congress Cataloging-in-Publication Data

Catalog information may be obtained from the Library of Congress

This book contains information obtained from authentic and highly regarded sources. Reprinted material is quotedwith permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been madeto publish reliable data and information, but the author and the publisher cannot assume responsibility for the validityof all materials or for the consequences of their use.

Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic ormechanical, including photocopying, microfilming, and recording, or by any information storage or retrieval system,without prior permission in writing from the publisher.

The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for creatingnew works, or for resale. Specific permission must be obtained in writing from CRC Press LLC for such copying.

Direct all inquiries to CRC Press LLC, 2000 Corporate Blvd., N.W., Boca Raton, Florida 33431.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used onlyfor identification and explanation, without any intent to infringe.

© 1999 by CRC Press LLC

No claim to original U.S. Government worksInternational Standard Book Number 0-8493-8008-1Printed in the United States of America 1 2 3 4 5 6 7 8 9 0Printed on acid-free paper

Table of Contents

PrefaceThe Author

Chapter 1 CONTEXT AND BACKGROUN D1.1. Emulsion Science in the Food Industry1.2. General Characteristics of Food Emulsions

1.2.1. Definitions1.2.2 Mechanisms of Emulsion Instability1.2.3. Ingredient Partitioning in Emulsions1.2.4. Dynamic Nature of Emulsions1.2.5. Complexity of Food Emulsions

1.3. Emulsion Properties1.3.1. Dispersed-Phase Volume Fraction1.3.2. Particle Size Distribution1.3.3. Interfacial Properties1.3.4. Droplet Charge1.3.5. Droplet Crystallinity

1.4. Hierarchy of Emulsion Properties1.5. Investigation of Emulsion Properties1.6. Overview and Philosophy

Chapter 2 MOLECULAR INTERACTIONS2.1. Introduction2.2. Forces of Nature2.3. Origin and Nature of Molecular Interactions

2.3.1. Covalent Interactions2.3.2. Electrostatic Interactions2.3.3. van der Waals Interactions2.3.4. Steric Overlap Interactions

2.4. Overall Intermolecular Pair Potential2.5. Bond Strengths and the Role of Thermal Energy2.6. The Structural Organization of Molecules in Liquids

2.6.1. Thermodynamics of Mixing2.6.2. Potential Energy Change on Mixing2.6.3. Entropy Change on Mixing2.6.4. Free Energy Change on Mixing2.6.5. The Properties of More Complex Systems

2.7. Molecular Interactions and Conformation2.8. Higher Order Interactions

2.8.1. Hydrogen Bonds2.8.2. Hydrophobic Interactions

©1999 CRC Press LLC

2.9. Computer Modeling of Liquid Properties2.9.1. Monte Carlo Techniques2.9.2. Molecular Dynamics Techniques

Chapter 3 COLLOIDAL INTERACTIONS3.1. Introduction3.2. Colloidal Interactions and Droplet Aggregation3.3. van der Waals Interactions

3.3.1. Origin of van der Waals Interactions3.3.2. Interdroplet Pair Potential3.3.3. Hamaker Function3.3.4. Electrostatic Screening3.3.5. Retardation3.3.6. Influence of an Interfacial Layer3.3.7. General Features of van der Waals Interactions

3.4. Electrostatic Interactions3.4.1. Origins of Surface Charge3.4.2. Ion Distribution Near a Charged Surface3.4.3. Electrostatic Interactions Between Charged Droplets3.4.4. Ion Bridging3.4.5. General Features of Electrostatic Interactions

3.5. Polymeric Steric Interactions3.5.1. Polymeric Emulsifiers3.5.2. Interdroplet Pair Potential3.5.3. Distance Dependence of Polymeric Steric Interactions3.5.4. Optimum Characteristics of Polymeric Emulsifier3.5.5. General Features of Polymeric Steric Stabilization

3.6. Depletion Interactions3.6.1. Origin of Depletion Interactions3.6.2. Interdroplet Pair Potential3.6.3. General Features of Depletion Interactions

3.7. Hydrophobic Interactions3.8. Hydration Interactions3.9. Thermal Fluctuation Interactions

3.9.1. Protrusion Interactions3.9.2. Undulation Interactions

3.10. Hydrodynamic Interactions and Nonequilibrium Effects3.11. Total Interaction Potential

3.11.1. van der Waals and Electrostatic3.11.2. van der Waals, Electrostatic, and Steric3.11.3. van der Waals and Steric3.11.4. van der Waals, Electrostatic, and Hydrophobic3.11.5. van der Waals, Electrostatic, and Depletion

3.12. Predicting Colloidal Interactions in Food Emulsions

Chapter 4 EMULSION INGREDIENTS4.1. Introduction4.2. Fats and Oils

4.2.1. Molecular Structure and Organization


4.2.2. Bulk Physicochemical Properties4.2.3. Fat Crystallization4.2.4. Chemical Changes

4.3. Water4.3.1. Molecular Structure and Organization4.3.2. Bulk Physicochemical Properties

4.4. Aqueous Solutions4.4.1. Interaction of Water with Ionic Solutes4.4.2. Interaction of Water with Dipolar Solutes4.4.3. Interaction of Water with Nonpolar Solutes: The Hydrophobic Effect

4.5. Surfactants4.5.1. Molecular Characteristics4.5.2. Functional Properties4.5.3. Surfactant Classification

4.6. Biopolymers4.6.1. Molecular Characteristics4.6.2. Molecular Basis of Conformation and Aggregation4.6.3. Functional Properties4.6.4. Modification of Biopolymers4.6.5. Ingredient Selection

Chapter 5 INTERFACIAL PROPERTIES AND THEIRCHARACTERIZATION

5.1. Introduction5.2. Molecular Basis of Interfacial Properties

5.2.1. Interfaces Between Two Pure Liquids5.2.2. Interfaces with Adsorbed Emulsifiers

5.3. Thermodynamics of Interfaces5.3.1. Gas–Liquid Interface in the Absence of an Emulsifier5.3.2. Gas–Liquid Interface in the Presence of an Emulsifier5.3.3. Liquid–Liquid Interface5.3.4. Measurement of the Surface Excess Concentration

5.4. Properties of Curved Interfaces5.5. Contact Angles and Wetting5.6. Capillary Rise and Meniscus Formation5.7. Adsorption Kinetics

5.7.1. Movement of Molecules to an Interface5.7.2. Incorporation of Emulsifier Molecules at an Interface

5.8. Common Types of Interfacial Membranes5.9. Interfacial Composition and Competitive Adsorption5.10. Measurement of Surface and Interfacial Tensions

5.10.1. Du Nouy Ring Method5.10.2. Wilhelmy Plate Method5.10.3. Sessile- and Pendant-Drop Methods5.10.4. Drop-Volume Method5.10.5. Spinning-Drop Method5.10.6. Maximum Bubble Pressure Method5.10.7. Oscillating Jet Method5.10.8. Capillary Wave Method


5.11. Interfacial Rheology5.11.1. Measurement of Interfacial Shear Rheology5.11.2. Measurement of Interfacial Dilatational Rheology

5.12. Characterization of Interfacial Structure5.13. Practical Implications of Interfacial Phenomena

Chapter 6 EMULSION FORMATION6.1. Introduction6.2. Overview of Homogenization6.3. Physical Principles of Emulsion Formation

6.3.1. Droplet Disruption6.3.2. Droplet Coalescence6.3.3. The Role of the Emulsifier

6.4. Homogenization Devices6.4.1. High-Speed Blenders6.4.2. Colloid Mills6.4.3. High-Pressure Valve Homogenizers6.4.4. Ultrasonic Homogenizers6.4.5. Microfluidization6.4.6. Membrane Homogenizers6.4.7. Homogenization Efficiency6.4.8. Comparison of Homogenizers

6.5. Factors Which Influence Droplet Size6.5.1. Emulsifier Type and Concentration6.5.2. Energy Input6.5.3. Properties of Component Phases6.5.4. Temperature

6.6. Demulsification6.6.1. Nonionic Surfactants6.6.2. Ionic Surfactants6.6.3. Biopolymer Emulsifiers6.6.4. General Methods of Demulsification6.6.5. Selection of Most Appropriate Demulsification Technique

6.7. Future Developments

Chapter 7 EMULSION STABILITY7.1. Introduction7.2. Energetics of Emulsion Stability

7.2.1. Thermodynamics7.2.2. Kinetics

7.3. Gravitational Separation7.3.1. Physical Basis of Gravitational Separation7.3.2. Methods of Controlling Gravitational Separation7.3.3. Experimental Measurement of Gravitational Separation

7.4. Flocculation7.4.1. Physical Basis of Flocculation7.4.2. Methods of Controlling Flocculation7.4.3. Structure and Properties of Flocculated Emulsions7.4.4. Experimental Measurement of Flocculation


7.5. Coalescence7.5.1. Physical Basis of Coalescence7.5.2. Methods of Controlling Coalescence7.5.3. Influence of Emulsifier Type and Environmental Conditions7.5.4. Experimental Measurement of Droplet Coalescence

7.6. Partial Coalescence7.6.1. Physical Basis of Partial Coalescence7.6.2. Methods of Controlling Partial Coalescence7.6.3. Experimental Measurement of Partial Coalescence

7.7. Ostwald Ripening7.7.1. Physical Basis of Ostwald Ripening7.7.2. Methods of Controlling Ostwald Ripening7.7.3. Experimental Measurement of Ostwald Ripening

7.8. Phase Inversion7.8.1. Physical Basis of Phase Inversion7.8.2. Methods of Controlling Phase Inversion7.8.3. Experimental Measurement of Phase Inversion

7.9. Chemical and Biochemical Stability

Chapter 8 EMULSION RHEOLOGY8.1. Introduction8.2. Rheological Properties of Materials

8.2.1. Solids8.2.2. Liquids8.2.3. Plastics8.2.4. Viscoelastic Materials

8.3. Measurement of Rheological Properties8.3.1. Simple Compression and Elongation8.3.2. Shear Measurements8.3.3. Empirical Techniques

8.4. Rheological Properties of Emulsions8.4.1. Dilute Suspensions of Rigid Spherical Particles8.4.2. Dilute Suspensions of Fluid Spherical Particles8.4.3. Dilute Suspensions of Rigid Nonspherical Particles8.4.4. Dilute Suspensions of Flocculated Particles8.4.5. Concentrated Suspensions of Nonflocculated Particles in the

Absence of Colloid Interactions8.4.6. Suspensions of Nonflocculated Particles with Repulsive Interactions8.4.7. Concentrated Suspensions of Flocculated Particles8.4.8. Emulsions with Semisolid Continuous Phases

8.5. Major Factors that Determine Emulsion Rheology8.5.1. Dispersed-Phase Volume Fraction8.5.2. Rheology of Component Phases8.5.3. Droplet Size8.5.4. Colloidal Interactions8.5.5. Particle Charge

Chapter 9 APPEARANCE AND FLAVOR9.1. Introduction9.2. Emulsion Flavor


9.2.1. Flavor Partitioning9.2.2. Flavor Release9.2.3. Measurements of Partition Coefficients and Flavor Release

9.3. Emulsion Appearance9.3.1. Interaction of Light Waves with Emulsions9.3.2. Overall Appearance9.3.3. Experimental Techniques

Chapter 10 CHARACTERIZATION OF EMULSION PROPERTIES10.1. Introduction

10.1.1. Research and Development10.1.2. Quality Control

10.2. Testing Emulsifier Efficiency10.2.1. Emulsifying Capacity10.2.2. Emulsion Stability Index10.2.3. Interfacial Tension10.2.4. Interfacial Rheology

10.3. Microstructure and Droplet Size Distribution10.3.1. Microscopy10.3.2. Static Light Scattering10.3.3. Dynamic Light Scattering10.3.4. Electrical Pulse Counting10.3.5. Sedimentation Techniques10.3.6. Ultrasonic Spectrometry10.3.7. Nuclear Magnetic Resonance10.3.8. Neutron Scattering10.3.9. Dielectric Spectroscopy10.3.10. Electroacoustics

10.4. Dispersed-Phase Volume Fraction10.4.1. Proximate Analysis10.4.2. Density Measurements10.4.3. Electrical Conductivity10.4.4. Alternative Techniques

10.5. Droplet Crystallinity10.5.1. Dilatometry10.5.2. Nuclear Magnetic Resonance10.5.3. Thermal Analysis10.5.4. Ultrasonics

10.6. Droplet Charge10.6.1. Electrophoresis10.6.2. Zetasizer©

10.6.3. Electroacoustics

REFERENCES


Preface

As one strolls along the aisles of a supermarket, one passes a wide variety of food products,both natural and manufactured, which exist either partly or wholly as emulsions or whichhave been in an emulsified form sometime during their production. Common examplesinclude milk, flavored milks, creams, salad dressings, dips, coffee whitener, ice cream, soups,sauces, mayonnaise, butter, margarine, fruit beverages, and whipped cream. Even thoughthese products differ widely in their appearance, texture, taste, and shelf life, they all consist(or once consisted) of small droplets of one liquid dispersed in another liquid. Consequently,many of their physicochemical and sensory properties can be understood by applying thefundamental principles and techniques of emulsion science. It is for this reason that anyonein the food industry working with these types of products should have at least an elementaryunderstanding of this important topic.

The primary objective of this book is to present the principles and techniques of emulsionscience and show how they can be used to better understand, predict, and control theproperties of a wide variety of food products. Rather than describe the specific methods andproblems associated with the creation of each particular type of emulsion-based food product,I have concentrated on an explanation of the basic concepts of emulsion science, as these areapplicable to all types of food emulsion. Details about the properties of particular types offood emulsion are described in the latest edition of an excellent book edited by S.E. Fribergand K. Larsson (Food Emulsions, 3rd edition, Marcel Dekker, New York, 1997), whichshould be seen as being complementary to this volume.

It is a great pleasure to acknowledge the contributions of all those who helped bring thisbook to fruition. Without the love and support of my best friend and partner Jayne and of myfamily, this book would never have been completed. I also thank all of my students and co-workers who have been a continual source of stimulating ideas and constructive criticism andmy teachers for providing me with the strong academic foundations on which I have at-tempted to build. Finally, I thank all those at CRC Press for their help in the preparation ofthis book.


The Author

Dr. David Julian McClements has been an Assistant Pro-fessor in the Department of Food Science at the Universityof Massachusetts since 1994. He received a B.S. (Hons) infood science (1985) and a Ph.D. in “Ultrasonic Character-ization of Fats and Emulsions” (1989) from the Universityof Leeds (United Kingdom). He then did postdoctoral re-search at the University of Leeds, University of Californiaat Davis, and the University College Cork in Ireland, be-fore joining the University of Massachusetts. Dr.McClements’ research interests include ultrasonic charac-terization of food emulsions, food biopolymers and col-loids (focusing on emulsions, gels, and micellar systems),protein functionality, and physicochemical properties oflipids.

Dr. McClements has co-authored a book entitled Ad-vances in Food Colloids with Professor Eric Dickinson and co-edited a book entitled Devel-opments in Acoustics and Ultrasonics with Dr. Malcolm Povey. In addition, he has publishedover 100 scientific articles as book chapters, encyclopedia entries, journal manuscripts, andconference proceedings. Dr. McClements recently received the Young Scientist award fromthe American Chemical Society’s Division of Food and Agriculture in recognition of hisachievements.


1 Context and Background

1.1. EMULSION SCIENCE IN THE FOOD INDUSTRY

Many natural and processed foods consist either partly or wholly as emulsions or have beenin an emulsified state at some time during their production; such foods include milk, cream,butter, margarine, fruit beverages, soups, cake batters, mayonnaise, cream liqueurs, sauces,desserts, salad cream, ice cream, and coffee whitener (Friberg and Larsson 1997, Krog et al.1983, Jaynes 1983, Dickinson and Stainsby 1982, Dickinson 1992, Swaisgood 1996). Emul-sion-based food products exhibit a wide variety of different physicochemical and organolep-tic characteristics, such as appearance, aroma, texture, taste, and shelf life. For example, milkis a low-viscosity white fluid, strawberry yogurt is a pink viscoelastic gel, and margarine isa yellow semisolid. This diversity is the result of the different sorts of ingredients andprocessing conditions used to create each type of product. The manufacture of an emulsion-based food product with specific quality attributes depends on the selection of the mostappropriate raw materials (e.g., water, oil, emulsifiers, thickening agents, minerals, acids,bases, vitamins, flavors, colorants, etc.) and processing conditions (e.g., mixing, homogeni-zation, pasteurization, sterilization, etc.).

Traditionally, the food industry largely relied on craft and tradition for the formulation offood products and the establishment of processing and storage conditions. This approach isunsuitable for the modern food industry, which must rapidly respond to changes in consumerpreferences for a greater variety of cheaper, healthier, and more convenient foods (Sloan1994, 1996; Katz 1997). In addition, the modern food industry relies increasingly on large-scale production operations to produce vast quantities of foods at relatively low cost. Thedevelopment of new foods, the improvement of existing foods, and the efficient running offood-processing operations require a more systematic and rigorous approach than was usedpreviously (Hollingsworth 1995).

Two areas which have been identified as being of particular importance to the improve-ment of food products are:

1. Enhanced scientific understanding of food properties. An improved understand-ing of the factors that determine the bulk physicochemical and organoleptic prop-erties of emulsions will enable manufacturers to create low-cost high-quality foodproducts in a more systematic and reliable fashion (Kokini et al. 1993, Rizvi et al.1993).

2. Development of new analytical techniques to characterize food properties. Thedevelopment and application of new analytical techniques to characterize theproperties of emulsions are leading to considerable advances in research, develop-ment, and quality control (Dickinson 1995a,b; Gaonkar 1995). These techniquesare used in the laboratory to enhance our understanding of the factors whichdetermine the properties of foods and in the factory to monitor the properties offoods during processing in order to ensure that they meet the required qualityspecifications.


Emulsion science is a multidisciplinary subject that combines chemistry, physics, andengineering (Sherman 1968a; Becher 1957, 1983; Hiemenz 1986; Hunter 1986, 1989, 1993;Evans and Wennerstrom 1994). The aim of the emulsion scientist working in the foodindustry is to utilize the principles and techniques of emulsion science to enhance the qualityof the food supply and the efficiency of food production. This book presents the conceptualand theoretical framework required by food scientists to understand and control the propertiesof emulsion-based food products.

1.2. GENERAL CHARACTERISTICS OF FOOD EMULSIONS

1.2.1. Definitions

An emulsion consists of two immiscible liquids (usually oil and water), with one of theliquids dispersed as small spherical droplets in the other (Figure 1.1). In most foods, thediameters of the droplets usually lie somewhere between 0.1 and 100 µm (Dickinson andStainsby 1982; Dickinson 1992; Walstra 1996a,b). Emulsions can be conveniently classifiedaccording to the distribution of the oil and aqueous phases. A system which consists of oildroplets dispersed in an aqueous phase is called an oil-in-water or O/W emulsion (e.g.,mayonnaise, milk, cream, soups, and sauces). A system which consists of water dropletsdispersed in an oil phase is called a water-in-oil or W/O emulsion (e.g., margarine, butter, and

FIGURE 1.1 Microscopic image of a 20 wt% tetradecane oil-in-water emulsion stabilized by 1 wt%whey protein isolate obtained using confocal scanning fluorescence microscopy. The dark regions arethe oil droplets, and the light regions are the aqueous phase.


spreads). The substance that makes up the droplets in an emulsion is referred to as thedispersed or internal phase, whereas the substance that makes up the surrounding liquid iscalled the continuous or external phase. It is also possible to prepare multiple emulsions ofthe oil-in-water-in-oil (O/W/O) or water-in-oil-in-water (W/O/W) type (Dickinson andMcClements 1995). For example, a W/O/W emulsion consists of water droplets dispersedwithin larger oil droplets, which are themselves dispersed in an aqueous continuous phase(Evison et al. 1995). Recently, research has been carried out to create stable multiple emul-sions which can be used to control the release of certain ingredients, reduce the total fatcontent of emulsion-based food products, or isolate one ingredient from another (Dickinsonand McClements 1995).

The concentration of droplets in an emulsion is usually described in terms of the dispersed-phase volume fraction (φ) (Section 1.3.1). The process of converting two separate immiscibleliquids into an emulsion, or of reducing the size of the droplets in a preexisting emulsion, isknown as homogenization. In the food industry, this process is usually carried out usingmechanical devices known as homogenizers, which subject the liquids to intense mechanicalagitation (Chapter 6).

It is possible to form an emulsion by homogenizing pure oil and pure water together, butthe two phases rapidly separate into a system which consists of a layer of oil (lower density)on top of a layer of water (higher density). This is because droplets tend to merge with theirneighbors when they collide with them, which eventually leads to complete phase separation.The driving force for this process is the fact that the contact between oil and water moleculesis energetically unfavorable (Israelachvili 1992), so that emulsions are thermodynamicallyunstable systems (Chapter 7). It is possible to form emulsions that are kinetically stable(metastable) for a reasonable period of time (a few days, weeks, months, or years) byincluding substances known as emulsifiers and/or thickening agents prior to homogenization(Chapter 4). Emulsifiers are surface-active molecules which absorb to the surface of freshlyformed droplets during homogenization, forming a protective membrane which prevents thedroplets from coming close enough together to aggregate (Chapters 6 and 7). Most emulsi-fiers are amphiphilic molecules (i.e., they have polar and nonpolar regions on the samemolecule). The most common emulsifiers used in the food industry are amphiphilic proteins,small-molecule surfactants, and phospholipids (Chapter 4). Thickening agents are ingredientswhich are used to increase the viscosity of the continuous phase of emulsions, and theyenhance emulsion stability by retarding the movement of the droplets. The most commonthickening agents used in the food industry are polysaccharides (Chapter 4). A stabilizer isany ingredient that can be used to enhance the stability of an emulsion and may therefore beeither an emulsifier or a thickening agent.

An appreciation of the difference between the thermodynamic stability of a system and itskinetic stability is crucial for an understanding of the properties of food emulsions (Dickinson1992). Consider a system which consists of a large number of molecules that can occupy twodifferent states: Elow and Ehigh (Figure 1.2). The state with the lowest free energy is the onewhich is thermodynamically favorable and therefore the one that the molecules are mostlikely to occupy. At thermodynamic equilibrium, the two states are populated according tothe Boltzmann distribution (Atkins 1994):

φφ

high

low

low high= −−

exp( )E E

kT(1.1)

where φ is the fraction of molecules that occupies the energy level E, k is Boltzmann’sconstant (k = 1.38 × 10–23 J K–1), and T is the absolute temperature. The larger the difference


between the two energy levels compared to the thermal energy of the system (kT), thegreater the fraction of molecules in the lower energy state. In practice, a system may not beable to reach equilibrium during the time scale of an observation because of the presenceof an energy barrier (∆E*) between the two states (Figure 1.2). A system in the high energystate must acquire an energy greater than ∆E* before it can move into the low energy state.The rate at which a transformation from a high to a low energy state occurs thereforedecreases as the height of the energy barrier increases. When the energy barrier is suffi-ciently large, the system may remain in a thermodynamically unstable state for a consider-able length of time, in which case it is said to be kinetically stable or metastable (Atkins1994). In food emulsions, there are actually a large number of intermediate metastable statesbetween the initial emulsion and the separated phases, and there is an energy barrierassociated with a transition between each of these states. Nevertheless, it is often possibleto identify a single energy barrier, which is associated with a particular physicochemicalprocess, that is the most important factor in determining the overall kinetic stability of anemulsion (Chapter 7).

1.2.2. Mechanisms of Emulsion Instability

The term “emulsion stability” is broadly used to describe the ability of an emulsion to resistchanges in its properties with time (Chapter 7). Nevertheless, there are a variety of physico-chemical mechanisms which may be responsible for alterations in the properties of anemulsion, and it is crucial to be clear about which of these mechanisms are important in thesystem under consideration. A number of the most important physical mechanisms respon-sible for the instability of emulsions are shown schematically in Figure 1.3. Creaming andsedimentation are both forms of gravitational separation. Creaming describes the upwardmovement of droplets due to the fact that they have a lower density than the surroundingliquid, whereas sedimentation describes the downward movement of droplets due to the factthat they have a higher density than the surrounding liquid. Flocculation and coalescence areboth types of droplet aggregation. Flocculation occurs when two or more droplets cometogether to form an aggregate in which the droplets retain their individual integrity, whereascoalescence is the process where two or more droplets merge together to form a single largerdroplet. Extensive droplet coalescence can eventually lead to the formation of a separate layerof oil on top of a sample, which is known as “oiling off.” Phase inversion is the processwhereby an oil-in-water emulsion is converted into a water-in-oil emulsion or vice versa. The

FIGURE 1.2 Difference between thermodynamic and kinetic stability. A system will remain in athermodynamically unstable or metastable state for some time if there is a sufficiently large energybarrier preventing it from reaching the state with the lowest free energy.


FIGURE 1.4 The ingredients in an emulsion partition themselves between the oil, water, and inter-facial regions according to their concentration and interactions with the local environment.

factors which determine these and the other major forms of emulsion instability are discussedin Chapter 7, along with methods of controlling and monitoring them. In addition to thephysical processes mentioned above, it should be noted that there are also various chemical,biochemical, and microbiological processes that occur in food emulsions which can alsoaffect their shelf life and quality.

1.2.3. Ingredient Partitioning in Emulsions

Most food emulsions can conveniently be considered to consist of three regions which havedifferent physicochemical properties: the interior of the droplets, the continuous phase, andthe interface (Figure 1.4). The molecules in an emulsion distribute themselves among thesethree regions according to their concentration and polarity (Wedzicha 1988). Nonpolarmolecules tend to be located primarily in the oil phase, polar molecules in the aqueousphase, and amphiphilic molecules at the interface. It should be noted that even at equilib-rium, there is a continuous exchange of molecules between the different regions, whichoccurs at a rate that depends on the mass transport of the molecules through the system.Molecules may also move from one region to another when there is some alteration in theenvironmental conditions of an emulsion (e.g., a change in temperature or dilution withinthe mouth). The location and mass transport of the molecules within an emulsion have asignificant influence on the aroma, flavor release, texture, and physicochemical stability offood products (Dickinson and Stainsby 1982, Wedzicha et al. 1991, Coupland and McClements1996, Landy et al. 1996).

FIGURE 1.3 Food emulsions may become unstable through a variety of physical mechanisms,including creaming, sedimentation, flocculation, coalescence, and phase inversion.


1.2.4. Dynamic Nature of Emulsions

Many of the properties of emulsions can only be understood with reference to their dynamicnature. The formation of emulsions by homogenization is a highly dynamic process whichinvolves the violent disruption of droplets and the rapid movement of surface-active mol-ecules from the bulk liquids to the interfacial region (Chapter 6). Even after their formation,the droplets in an emulsion are in continual motion and frequently collide with one anotherbecause of their Brownian motion, gravity, or applied mechanical forces (Melik and Fogler1988, Dukhin and Sjoblom 1996, Lips et al. 1993). The continual movement and interactionsof droplets cause the properties of emulsions to evolve over time due to the various desta-bilization mechanisms mentioned in Section 1.2.2. An appreciation of the dynamic processesthat occur in food emulsions is therefore extremely important for a thorough understandingof their bulk physicochemical and organoleptic properties.

1.2.5. Complexity of Food Emulsions

Most food emulsions are much more complex than the simple three-component (oil, water,and emulsifier) systems described in Section 1.2.1. The aqueous phase may contain a varietyof water-soluble ingredients, including sugars, salts, acids, bases, surfactants, proteins, andcarbohydrates. The oil phase usually contains a complex mixture of lipid-soluble compo-nents, such as triacylglycerols, diacylglycerols, monoacylglycerols, free fatty acids, sterols,and vitamins. The interfacial region may contain a mixture of various surface-active compo-nents, including proteins, phospholipids, surfactants, alcohols, and solid particles. In addition,these components may form various types of structural entities in the oil, water, or interfacialregions, such as fat crystals, ice crystals, protein aggregates, air bubbles, liquid crystals, andsurfactant micelles. A further complicating factor is that foods are subjected to variations intheir temperature, pressure, and mechanical agitation during their production, storage, andhandling, which can cause significant alterations in their overall properties.

It is clear from the above discussion that food emulsions are compositionally, structurally,and dynamically complex materials and that many factors contribute to their overall proper-ties. One of the major objectives of this book is to present the conceptual framework neededby food scientists to understand these complex systems in a more systematic and rigorousfashion. Much of our knowledge about these complex systems has come from studies ofsimple model systems (Section 1.5). Nevertheless, there is an increasing awareness of theneed to elucidate the factors that determine the properties of actual emulsion-based foodproducts. For this reason, many researchers are now focusing on the complex issues that needto be addressed, such as ingredient interactions, effects of processing conditions, and phasetransitions (Dickinson 1992, 1995b; Dickinson and McClements 1995; Dalgleish 1996a;Hunt and Dalgleish 1994, 1995; Demetriades et al. 1997a,b).

1.3. EMULSION PROPERTIES

1.3.1. Dispersed-Phase Volume Fraction

The concentration of droplets in an emulsion is usually described in terms of the dispersed-phase volume fraction (φ), which is equal to the volume of emulsion droplets (VD) dividedby the total volume of the emulsion (VE): φ = VD/VE. Knowledge of the dispersed-phasevolume fraction is important because the droplet concentration influences the appearance,texture, flavor, stability, and cost of emulsion-based food products. In some situations, it ismore convenient to express the composition of an emulsion in terms of the dispersed-phasemass fraction (φm), which is related to the volume fraction by the following equation:


φφρ

ρ φ φ ρm =+ −

2

2 11( ) (1.2)

where ρ1 and ρ2 are the densities of the continuous and dispersed phases, respectively. Whenthe densities of the two phases are equal, the mass fraction is equivalent to the volumefraction. The dispersed-phase volume fraction of an emulsion is often known because theconcentration of the ingredients used to prepare it is carefully controlled. Nevertheless, localvariations in dispersed-phase volume fraction occur within emulsions when the dropletsaccumulate at either the top or bottom of an emulsion due to creaming or sedimentation. Inaddition, the dispersed-phase volume fraction of an emulsion may vary during a food-processing operation (e.g., if a mixer or valve is not operating efficiently). Consequently, itis important to have analytical techniques to measure dispersed-phase volume fraction(Chapter 10).

1.3.2. Particle Size Distribution

Many of the most important properties of emulsion-based food products (e.g., shelf life,appearance, texture, and flavor) are determined by the size of the droplets they contain,(Dickinson and Stainsby 1982, Dickinson 1992). Consequently, it is important for foodscientists to be able to reliably control, predict, measure, and report the size of the dropletsin emulsions. In this section, the most important methods of reporting droplet sizes arediscussed. Methods of controlling, predicting, and measuring droplet size are covered in laterchapters (Chapters 6, 7, and 10).

If all the droplets in an emulsion are of the same size, the emulsion is referred to asmonodisperse, but if there is a range of sizes present, the emulsion is referred to as polydis-perse. The size of the droplets in a monodisperse emulsion can be completely characterizedby a single number, such as the droplet diameter (d) or radius (r). Monodisperse emulsionsare sometimes used for fundamental studies because the interpretation of experimentalmeasurements is much simpler than that of polydisperse emulsions. Nevertheless, foodemulsions always contain a distribution of droplet sizes, and so the specification of theirdroplet size is more complicated than that of monodisperse systems. Ideally, one wouldlike to have information about the full particle size distribution of an emulsion (i.e., thesize of each of the droplets in the system). Nevertheless, in many situations, knowledgeof the average size of the droplets and the width of the distribution is sufficient (Hunter1986).

1.3.2.1. Presenting Particle Size Data

The number of droplets in most emulsions is extremely large, and so their size can beconsidered to vary continuously from some minimum value to some maximum value. Whenpresenting particle size data, it is convenient to divide this size range into a number ofdiscrete size classes and stipulate the number of droplets that fall into each class (Hunter1986). The resulting data can then be represented in tabular form (Table 1.1) or plotted asa histogram that shows the number of droplets in each size class (Figure 1.5). Rather thanpresenting the number of droplets (ni) in each size class, it is often more informative topresent the data as the number frequency, fi = ni /N, where N is the total number of droplets,or as the volume frequency, φi = vi /V, where vi is the volume of the droplets in the ith sizeclass and V is the total volume of all the droplets in the emulsion. It should be noted thatthe shape of a particle size distribution changes appreciably depending on whether it ispresented as a number or volume frequency (Table 1.2). The volume of a droplet is propor-


TABLE 1.1The Particle Size Distribution of an Emulsion Represented in Tabular Form

Size class di fi i C(di)(m) (m) Ni (%) (%) (%)

041–0.054 0.048 0 0.0 0.0 0.00.054–0.071 0.063 2 0.1 0.0 0.10.071–0.094 0.082 4 0.2 0.0 0.30.094–0.123 0.108 50 2.5 0.0 2.80.123–0.161 0.142 84 4.2 0.1 7.00.161–0.211 0.186 152 7.6 0.3 14.60.211–0.277 0.244 224 11.2 1.1 25.80.277–0.364 0.320 351 17.6 3.9 43.350.364–0.477 0.420 470 23.5 11.8 66.850.477–0.626 0.551 385 19.2 21.8 86.10.626–0.821 0.723 190 9.5 24.3 95.60.821–1.077 0.949 64 3.2 18.5 98.81.077–1.414 1.245 21 1.0 13.7 99.851.414–1.855 1.634 3 0.2 4.4 1001.855–2.433 2.144 0 0.0 0.0 100

Note: The volume frequency is much more sensitive to larger droplets than is the number frequency.

tional to d3, and so a volume distribution is skewed more toward the larger droplets, whereasa number distribution is skewed more toward the smaller droplets.

A particle size distribution can also be represented as a smooth curve, such as the distri-bution function, F(di), or the cumulative function, C(di) (Figure 1.5). The (number) distribu-tion function is constructed so that the area under the curve between two droplet sizes (di anddi + δdi) is equal to the number of droplets (ni) in that size range (i.e., ni = F(di)δdi) (Hunter1986). This relationship can be used to convert a histogram to a distribution function or viceversa. The cumulative function represents the percentage of droplets that are smaller than di

(Figure 1.5). The resulting curve has an S-shape which varies from 0 to 100% as the particlesize increases. The particle size at which half the droplets are smaller and the other half arelarger is known as the median droplet diameter (dm).

1.3.2.2. Mean and Standard Deviation

It is often convenient to represent the size of the droplets in a polydisperse emulsion by oneor two numbers, rather than stipulating the full particle size distribution (Hunter 1986). Themost useful numbers are the mean diameter (d), which is a measure of the central tendencyof the distribution, and the standard deviation (σ), which is a measure of the width of thedistribution:

d n d N n d d Ni i i i= = −[ ]∑ ∑/ ( ) /σ 2 (1.3)

The above mean is also referred to as the mean length diameter (dL) because it representsthe sum of the length of the droplets divided by the total number of droplets. If all the dropletsin a polydisperse emulsion were laid end to end, they would have the same overall length asthose in a monodisperse emulsion containing an equal number of droplets of diameter dL. It


is also possible to express the mean droplet size in a number of other ways (Table 1.2). Eachof these mean sizes has dimensions of length (meters), but stresses a different physical aspectof the distribution (e.g., the average length, surface area, or volume). For example, thevolume–surface mean diameter is related to the surface area of droplets exposed to thecontinuous phase per unit volume of emulsion (AS):

AdS

VS

= 6φ(1.4)

This relationship is particularly useful for calculating the total surface area of droplets inan emulsion from a knowledge of the mean diameter of the droplets and the dispersed-phasevolume fraction. An appreciation of the various types of mean droplet diameter is alsoimportant because different experimental techniques used to measure droplet sizes aresensitive to different mean values (Orr 1988). For example, analysis of polydisperse emul-sions using osmotic pressure measurements gives information about their mean lengthdiameter, whereas light-scattering and sedimentation measurements give information abouttheir mean surface diameter. Consequently, it is always important to be clear about which

FIGURE 1.5 The particle size distribution of an emulsion can be represented by a histogram, adistribution function F(d), or a cumulative function C(d).


TABLE 1.2Different Ways of Expressing the Mean Droplet Diameterof a Polydisperse Emulsion

Name of mean Symbol Definition

Length d or dL d n d nL i i i= ∑∑ /

Surface area dS d n d nS i i i= ∑∑ 2 /

Volume dV d n d nV i i i= ∑∑ 33/

Volume–surface area dVS or d32 d n d n dVS i i i i= ∑∑ 3 2/

mean diameter has been determined in an experiment when using or quoting droplet sizedata.

1.3.2.3. Mathematical Models

The particle size distribution of an emulsion can often be modeled using a mathematicaltheory, which is convenient because it means that the full data set can be described by a smallnumber of parameters (Hunter 1986). If a plot of droplet frequency versus droplet size issymmetrical about the mean droplet size, the curve can be described by a normal distributionfunction (Figure 1.6):

f dd d

( ) exp( )= − −

1

2 2

2

2σ π σ (1.5)

where f(d)δd is the fraction of emulsion droplets which lies within the size interval betweend and d + δd. The number of droplets in each size group can be calculated from the relationni = N · f(d) · δd. Most (~68%) of the droplets fall within one standard deviation of the mean(d–

± σ), while the vast majority (∼ 99.7%) fall within three standard deviations (d– ± 3σ). Only

two parameters are needed to describe the particle size distribution of an emulsion that canbe approximated by a normal distribution, the mean and the standard deviation.

The particle size distribution of most food emulsions is not symmetrical about the mean,but tends to extend much further at the high-droplet-size end than at the low-droplet-size end(Figure 1.6). This type of distribution can often be described by a log-normal distribution:

f dd d

g

g

g

( )ln

exp(ln ln )

ln=

− −

1

2 2

2

2σ π σ (1.6)

where f(d)δ(ln d) is the fraction of emulsion droplets which lies within the size intervalbetween lnd and lnd + δ(ln d), and dg and σg are the geometric mean and the standarddeviation of the geometric mean, which are given by the following expressions:

ln ln / ln [ (ln ln ) ] /d n d N n d d Ng i i g i i g= = −∑∑ σ 2 (1.7)


If the log-normal curve shown in Figure 1.6 was plotted as droplet frequency versus logdiameter, it would be symmetrical about ln d

–g.

It should be stressed that the particle size distribution of many food emulsions cannot beadequately described by the simple models given above. Bimodal distributions, which arecharacterized by two peaks (Figure 1.7), are often encountered in food emulsions (e.g., whenextensive droplet flocculation occurs or when there is insufficient emulsifier present in anemulsion to stabilize all of the droplets formed during homogenization). For these systems,it is often better to present the data as the full particle size distribution; otherwise, consider-able errors may occur if an inappropriate model is used.

1.3.3. Interfacial Properties

The droplet interface consists of a narrow region (usually a few nanometers thick) whichsurrounds each emulsion droplet and contains a mixture of oil, water, and emulsifier mol-

FIGURE 1.6 Comparison of emulsions that can be described by normal and log-normal droplet sizedistributions.


ecules (Hunter 1986, 1989). The interfacial region only makes up a significant fraction of thetotal volume of an emulsion when the droplet size is less than about 1 µm (Table 1.3). Evenso, it plays a major role in determining many of the most important bulk physicochemical andorganoleptic properties of food emulsions. For this reason, food scientists are particularlyinterested in elucidating the factors which determine the composition, structure, thickness,rheology, and charge of the interfacial region. The composition and structure of the interfacialregion are determined by the type and concentration of surface-active species present, as wellas by the events which occur both during and after emulsion formation (Chapter 6). Thethickness and rheology of the interfacial region influence the stability of emulsions togravitational separation, coalescence, and flocculation and determine the rate at which mol-ecules leave or enter the droplets. The major factors which determine the characteristics ofthe interfacial region are discussed in Chapter 5, along with experimental techniques tocharacterize its properties.

TABLE 1.3Effect of Particle Size on the Physical Characteristics of 1 g of OilDispersed in Water in the Form of Spherical Droplets

No. of droplets Droplet surfaceDroplet radius per gram oil area per gram oil % oil molecules

(m) (g–1) (m2 g–1) at droplet surface

100 2.6 × 105 0.03 0.0210 2.6 × 108 0.3 0.21 2.6 × 1011 3 1.80.1 2.6 × 1014 30 18

Note: Values were calculated assuming the oil had a density of 920 kg m–3 and the end-to-end length ofthe oil molecules was 6 nm.

FIGURE 1.7 Example of a bimodal distribution resulting from the heat-induced flocculation ofdroplets in a 20 wt% corn oil-in-water emulsion stabilized by whey protein isolate. (From Demetriades,K., Coupland, J.N., and McClements, D.J., Journal of Food Science, 62, 462, 1997b. With permission.)


1.3.4. Droplet Charge

The bulk physicochemical and organoleptic properties of many food emulsions are governedby the magnitude and sign of the electrical charge on the droplets (Dickinson and Stainsby1982). The origin of this charge is normally the adsorption of emulsifier molecules that areionized or ionizable. Surfactants have hydrophilic head groups that may be neutral, positivelycharged, or negatively charged (Chapter 4). Proteins may also be neutral, positively charged,or negatively charged depending on the pH of the solution compared to their isoelectric point(Chapter 4). Consequently, emulsion droplets may have an electrical charge that depends onthe types of surface-active molecules present and the pH of the aqueous phase. The chargeon a droplet is important because it determines the nature of its interactions with othercharged species (Chapters 2 and 3) or its behavior in the presence of an electrical field(Chapter 10). Two species which have charges of opposite sign are attracted toward eachother, whereas two species which have charges of similar sign are repelled (Chapters 2 and3). All of the droplets in an emulsion are usually coated with the same type of emulsifier, andso they have the same electrical charge (if the emulsifier is ionized). When this charge issufficiently large, the droplets are prevented from aggregating because of the electrostaticrepulsion between them (Chapter 3). The properties of emulsions stabilized by ionizedemulsifiers are particularly sensitive to the pH and ionic strength of the aqueous phase. If thepH of the aqueous phase is adjusted so that the emulsifier loses its charge, or if salt is addedto “screen” the electrostatic interactions between the droplets, the repulsive forces may nolonger be strong enough to prevent the droplets from aggregating. Droplet aggregation oftenleads to a large increase in emulsion viscosity (Chapter 8) and may cause the droplets tocream more rapidly (Chapter 7).

The influence of electrostatic interactions on the stability of emulsions can clearly bedemonstrated if one adds a few drops of lemon juice to a glass of homogenized milk. Aftera few minutes, the milk changes from a low-viscosity emulsion containing isolated oildroplets to a viscous coagulum containing extensively flocculated droplets. This is becausethe lemon juice decreases the pH of the emulsion toward the isoelectric point of theproteins, thereby reducing the electrostatic repulsion between the droplets and causing dropletaggregation.

Electrostatic interactions also influence the interactions between emulsion droplets andother charged species, such as biopolymers, surfactants, vitamins, antioxidants, flavors, andminerals (Dickinson 1992, Landy et al. 1996, Coupland and McClements 1996, Mei et al.1998). These interactions often have significant implications for the overall quality of anemulsion product. For example, the volatility of a flavor is reduced when it is electrostaticallyattracted to the surface of an emulsion droplet, which alters the flavor profile of a food (Landyet al. 1996), or the susceptibility of oil droplets to lipid oxidation depends on whether thecatalyst is electrostatically attracted to the droplet surface (Mei et al. 1998). The accumulationof charged species at a droplet surface and the rate at which this accumulation takes placedepend on the sign of the charge of the species relative to that of the surface, the strength ofthe electrostatic interaction, the concentration of the species, and the presence of any othercharged species that might compete for the surface.

The above discussion highlights the importance of droplet charge in determining both thephysical and chemical properties of food emulsions. It is therefore important for food scien-tists to be able to predict, control, and measure droplet charge. For most food emulsions, itis difficult to accurately predict droplet charge because of the complexity of their compositionand the lack of suitable theories. Nevertheless, there is a fairly good understanding of themajor factors which influence droplet charge (Chapter 3) and of the effect of droplet charge


on the stability and rheology of emulsions (Chapters 7 and 8). In addition, a variety ofexperimental techniques have been developed to measure the magnitude and sign of thecharge on emulsion droplets (Chapter 10).

1.3.5. Droplet Crystallinity

The physical state of the droplets in an emulsion can influence a number of its most importantbulk physicochemical and organoleptic properties, including appearance, rheology, flavor,aroma, and stability (Dickinson and McClements 1995; Boode 1992; Boode and Walstra1993a,b; Walstra 1996b). The production of margarine and butter depends on a controlleddestabilization of an oil-in-water emulsion containing partly crystalline droplets. The stabilityof cream to shear and temperature cycling depends on the crystallization of the milk fatdroplets. The rate at which milk fat droplets cream depends on their density, which isdetermined by the fraction of the droplet which is solidified. The cooling sensation thatoccurs when fat crystals melt in the mouth contributes to the characteristic mouthfeel of manyfood products (Walstra 1987). A knowledge of the factors that determine the crystallizationand melting of emulsified substances, and of the effect that droplet-phase transitions have onthe properties of emulsions, is therefore particularly important to food scientists.* In oil-in-water emulsions, we are concerned with phase transitions of emulsified fat, whereas in water-in-oil emulsions, we are concerned with phase transitions of emulsified water. In the foodindustry, we are primarily concerned with the crystallization and melting of emulsified fats,because these transitions occur at temperatures that are commonly encountered during theproduction, storage, or handling of oil-in-water emulsions and because they usually have apronounced influence on the bulk properties of food emulsions. In contrast, phase transitionsof emulsified water are less likely to occur in foods because of the high degree of supercool-ing required to initiate crystallization (Clausse 1985).

The percentage of total fat in a sample which is solidified at a particular temperature isknown as the solid fat content (SFC). The SFC varies from 100% at low temperatures wherethe fat is completely solid to 0% at high temperatures where the fat is completely liquid. Theprecise nature of the SFC–temperature curve is an important consideration when selecting afat for a particular food product. The shape of this curve depends on the composition of thefat, the thermal and shear history of the sample, whether the sample is heated or cooled, theheating or cooling rate, the size of the emulsion droplets, and the type of emulsifier. Themelting and crystallization behavior of emulsified substances can be quite different from thatof the same substance in bulk (Dickinson and McClements 1995). The crystallization of bulkfats is considered in Chapter 4, while the additional factors that influence the crystallizationof emulsified fats are considered in Chapter 7. Experimental techniques that are used toprovide information about the crystallization and melting of emulsion droplets are describedin Chapter 10.

1.4. HIERARCHY OF EMULSION PROPERTIES

The bulk physicochemical and organoleptic properties of emulsion-based food products areultimately determined by the concentration, dimensions, interactions, and dynamics of thevarious types of structural entities present within them (e.g., atoms, molecules, molecularaggregates, crystals, micelles, droplets, air bubbles, and individual phases) (Figure 1.8). The

* It should be noted that the continuous phase of an emulsion is also capable of melting or crystallizing, which canhave a profound influence on the overall properties. For example, the characteristic texture of ice cream is partlydue to the presence of ice crystals in the aqueous continuous phase, whereas the rheology of butter and margarineis determined by the existence of a network of aggregated fat crystals in the oil continuous phase.


properties of emulsions can therefore be studied at a number of different levels of structuralorganization (e.g., subatomic, atomic, molecular, supramolecular, colloidal, microscopic,macroscopic, and organoleptic) depending on the concerns of the investigator (Eads 1994).Subatomic particles interact with each other via strong and weak nuclear forces to formatoms. Atoms interact with each other via covalent and ionic bonds to form molecules. Atomsand molecules interact with each other via various covalent and noncovalent forces to formseparate phases (which may be gas, liquid, or solid), simple solutions, molecular aggregates,and colloidal particles (Chapter 2). The bulk physicochemical and organoleptic properties ofan emulsion depend on the way these structural entities interact with one another to form theemulsion droplets, interfacial region, and continuous phase. A more complete understandingof the factors that determine the properties of emulsions depends on establishing the mostimportant processes that operate at each level of structural organization and linking thedifferent levels together. This is an extremely ambitious and complicated task that requiresmany years of painstaking research. Nevertheless, the knowledge gained from such anendeavor will enable food manufacturers to design and produce higher quality foods in amore cost-effective and systematic fashion. For this reason, the connection between molecu-lar, colloidal, and bulk physicochemical properties of food emulsions will be stressed throughoutthis book.

1.5. INVESTIGATION OF EMULSION PROPERTIES

Our understanding of the factors which determine the properties of food emulsions developsthrough a synthesis of experimentation and theory development. An investigator usuallystudies a particular aspect of a system by carefully designing and carrying out an experiment.The investigator then uses his or her knowledge to postulate a hypothesis to account for theobserved behavior, which is then tested and refined by further experimentation. Eventually,

FIGURE 1.8 Typical dimensions of structural entities commonly found in emulsion-based foodproducts.


a theory may be developed which enables one to better understand and predict the behaviorof the system. As further studies are carried out, this theory evolves with time and may evenbe replaced by a competing theory that better accounts for the observed behavior.

Food emulsions are extremely complex systems, and many factors operate in concert todetermine their overall properties. For this reason, experiments are usually carried out usingsimplified model systems which retain the essential features of the real system, but whichignore many of the secondary effects. For example, the emulsifying properties of proteins areoften investigated by using an isolated individual protein, pure oil, and pure water (Dickinson1992). In reality, a protein ingredient used in the food industry consists of a mixture ofdifferent proteins, sugars, salts, fats, and minerals, and the oil and aqueous phases maycontain a variety of different chemical constituents (Section 1.2.5). Nevertheless, by using awell-characterized model system, it is possible to elucidate the primary factors which influ-ence the properties of proteins in emulsions in a more quantitative fashion. Once theseprimary factors have been established, it is possible to increase the complexity of the modelby introducing additional variables and systematically examining their influence on theoverall properties. This incremental approach eventually leads to a thorough understandingof the factors that determine the properties of actual food emulsions and to the developmentof theories which can be used to describe and predict their behavior.

1.6. OVERVIEW AND PHILOSOPHY

It is impossible to cover every aspect of food emulsions in a book of this size. Of necessity,one must be selective about the material presented and the style in which it is presented.Rather than reviewing the practical knowledge associated with each particular type of emul-sion-based food product, the focus here will be on the fundamental principles of emulsionscience as applied to food systems because these principles are generally applicable to alltypes of food emulsion. Even so, real food emulsions will be used as examples where possiblein order to emphasize the practical importance of the fundamental approach. As mentionedearlier, particular attention will be paid to the relationship among molecular, colloidal, andbulk physicochemical properties of food emulsions, because this approach leads to the mostcomplete understanding of their behavior.

Throughout this book, it will be necessary to introduce a number of theories which havebeen developed to describe the properties of emulsions. Rather than concentrating on themathematical derivation of these theories, their physical significance will be highlighted, witha focus on their relevance to food scientists. A feeling for the major factors which determinethe properties of food emulsions can often be gained by programming these theories onto apersonal computer and systematically examining the role that each physical parameter playsin the equation.


2 Molecular Interactions

2.1. INTRODUCTION

Although food scientists have some control over the final properties of a product, they mustwork within the physical constraints set by nature (i.e., the characteristics of the individualmolecules and the type of interactions that occur between them). There is an increasingawareness within the food industry that the efficient production of foods with improvedquality depends on a better understanding of the molecular basis of their bulk physicochemi-cal and organoleptic properties (Baianu 1992, Kokini et al. 1993, Eads 1994). The individualmolecules within a food emulsion can interact with each other to form a variety of differentstructural entities (Figure 2.1). A molecule may be part of a bulk phase where it is surroundedby molecules of the same type, it may be part of a mixture where it is surrounded bymolecules of a different type, it may be part of an electrolyte solution where it is surroundedby counterions and solvent molecules, it may accumulate at an interface between two phases,it may be part of a molecular aggregate dispersed in a bulk phase, it may be part of a three-dimensional network that extends throughout the system, or it may form part of a complexbiological structure (Israelachvili 1992). The bulk physicochemical properties of food emul-sions depend on the nature, properties, and interactions of the structures formed by themolecules. The structural organization of a particular set of molecules is largely determinedby the forces that act between them and the prevailing environmental conditions (e.g.,temperature and pressure). Nevertheless, foods are rarely in their most thermodynamicallystable state, and therefore the structural organization of the molecules is often governed byvarious kinetic factors which prevent them from reaching the arrangement with the lowestfree energy (Section 1.2.1). For this reason, the structural organization of the molecules infoods is largely dependent on their previous history (i.e., the temperatures, pressures, gravity,and applied mechanical forces experienced during their lifetime). To understand, predict, andcontrol the behavior of food emulsions, it is important to be aware of the origin and natureof the forces responsible for holding the molecules together and how these forces lead to thevarious types of structures found in food emulsions. Only then will it be possible to createand stabilize foods that have internal structures that are known to be beneficial to foodquality.

2.2. FORCES OF NATURE

There are four distinct types of force in nature: strong nuclear interactions, weak nuclearinteractions, electromagnetic interactions, and gravity (Israelachvili 1992, Atkins 1994). Thestrong and weak nuclear forces act over extremely short distances and are chiefly responsiblefor holding together subatomic particles in the nucleus. As nuclear rearrangements do notnormally occur in foods, these forces will not be considered further. Gravitational forces arerelatively weak and act over large distances compared to other types of forces. Their strengthis proportional to the product of the masses of the objects involved, and consequently they


FIGURE 2.1 The molecules in food emulsions may adopt a variety of different structural arrange-ments depending on the nature of their interactions with their neighbors.

are insignificant at the molecular level because molecular masses are extremely small.Nevertheless, they do affect the behavior of food emulsions at the macroscopic level (e.g.,sedimentation or creaming of droplets, the shape adopted by large droplets, meniscus forma-tion, and capillary rise) (Israelachvili 1992). The forces that act at the molecular level are allelectromagnetic in origin and can conveniently be divided into four types: covalent, electro-static, van der Waals, and steric overlap (Hiemenz 1986, Israelachvili 1992, Atkins 1994).Despite acting over extremely short distances, often on the order of a few angstroms or less,intermolecular forces are ultimately responsible for the bulk physicochemical and organolep-tic properties of emulsions and other food materials.

2.3. ORIGIN AND NATURE OF MOLECULAR INTERACTIONS

2.3.1. Covalent Interactions

Covalent bonds involve the sharing of outer-shell electrons between two or more atoms, sothat the individual atoms lose their discrete nature (Karplus and Porter 1970, Atkins 1994).The number of electrons in the outer shell of an atom governs its valency (i.e., the optimumnumber of covalent bonds it can form with other atoms). Covalent bonds may be saturatedor unsaturated, depending on the number of electrons involved. Unsaturated bonds tend tobe shorter, stronger, and more rigid than saturated bonds (Israelachvili 1992). The distributionof the electrons within a covalent bond determines its polarity. When the electrons are sharedequally among the atoms, the bond has a nonpolar character, but when the electrons areshared unequally, the bond has a polar character. The polarity of a molecule depends on thesymmetry of the various covalent bonds which it contains (see Section 2.3.2.). Covalentbonds are also characterized by their directionality (i.e., their tendency to be directed atclearly defined angles relative to each other). The valency, saturation, polarity, strength, anddirectionality of covalent bonds determine the three-dimensional structure, flexibility, chemi-cal reactivity, and physical interactions of molecules.

Chemical reactions involve the breaking and formation of covalent bonds (Atkins 1994).The bulk physicochemical and organoleptic properties of food emulsions are altered byvarious types of chemical and biochemical reactions that occur during their production,storage, and consumption (Coultate 1996; Fennema 1996; Fennema and Tannenbaum 1996a,b).Some of these reactions are beneficial to food quality, while others are detrimental. It is


therefore important for food scientists to be aware of the various types of chemical reactionthat occur in food emulsions and to establish their influence on the overall properties of thesystem. The chemical reactions which occur in food emulsions are similar to those that occurin any other multicomponent heterogeneous food materials (e.g., oxidation of lipids [Nawar1996], hydrolysis of proteins or polysaccharides [Damodaran 1996, BeMiller and Whistler1996], cross-linking of proteins [Damodaran 1996], and Maillard reactions between reducingsugars and free amino groups [BeMiller and Whistler 1996]). Nevertheless, the rates andpathways of these reactions are often influenced by the physical environment of the mol-ecules involved (e.g., whether they are located in the oil, water, or interfacial region) (Wedzicha1988).

Until fairly recently, emulsion scientists were principally concerned with understanding thephysical changes which occur in food emulsions, rather than the chemical changes. Never-theless, there is currently great interest in establishing the relationship between emulsionproperties and the mechanisms of various chemical reactions that occur within them (Wedzichaet al. 1991, Coupland and McClements 1996, Landy et al. 1996, Huang et al. 1997).

Despite the importance of chemical reactions in emulsion quality, it should be stressedthat many of the most important changes in emulsion properties are a result of alterationsin the spatial distribution of the molecules, rather than the result of alterations in theirchemical structure (e.g., creaming, flocculation, coalescence, and phase inversion). Thespatial distribution of molecules is governed principally by their noncovalent (or physical)interactions with their neighbors (e.g., electrostatic, van der Waals, and steric overlap). It istherefore particularly important to have a good understanding of the origin and nature ofthese interactions.

2.3.2. Electrostatic Interactions

Electrostatic interactions occur between molecular species that possess a permanent electricalcharge, such as ions and polar molecules (Murrell and Boucher 1982, Reichardt 1988, Rogers1989). An ion is an atom or molecule that has either lost or gained one or more outer-shellelectrons so that it obtains a permanent positive or negative charge (Atkins 1994) (Figure2.2). A polar molecule has no net charge (i.e., as a whole, the molecule is neutral), but it doeshave an electrical dipole because of an uneven distribution of the charges within it. Certainatoms are able to “pull” the electrons in the covalent bonds toward them more strongly thanare other atoms (Atkins 1994). As a consequence, they acquire a partial negative charge(δ–), and the other atom acquires a partial positive charge (δ+). If the partial charges withina molecule are distributed symmetrically, they cancel each other and the molecule has nodipole (e.g., CCl4), but if they are distributed asymmetrically, the molecule will have a dipole

FIGURE 2.2 Schematic representation of the most important types of intermolecular electrostaticinteractions that arise between molecules.


(Israelachvili 1992). For example, the chlorine atom in HCl pulls the electrons in the covalentbond more strongly than the hydrogen atom, and so a dipole is formed: Hδ+Clδ–. The strengthof a dipole is characterized by the dipole moment µ = ql, where l is the distance between twocharges q+ and q–. The greater the magnitude of the partial charges, or the farther they areapart, the greater the dipole moment of a molecule.

The interaction between two molecular species is characterized by an intermolecular pairpotential, w(s), which is the energy required to bring two molecules from an infinite distanceapart to a separation s (Israelachvili 1992). There are a number of different types of electro-static interactions that can occur between permanently charged molecular species (ion–ion,ion–dipole, and dipole–dipole), but they can all be described by a similar equation (Hiemenz1986):

w sQ Q

sER

n( ) = 1 2

04πε ε (2.1)

where Q1 and Q2 are the effective charges on the two species, ε0 is the dielectric constant ofa vacuum (8.85 × 10–12 C2 J–1 m–1), εR is the relative dielectric constant of the interveningmedium, s is the center-to-center distance between the charges, and n is an integer thatdepends on the nature of the interaction. For ions, the value of Q is determined by theirvalency (z) and electrical charge (e) (1.602 × 10–19 C), whereas for dipoles, it is determinedby their dipole moment µ and orientation (Table 2.1). Numerical calculations of the intermo-lecular pair potential for representative ion–ion, ion–dipole, and dipole–dipole interactionsare illustrated in Figure 2.3a.

Examination of Equation 2.1 and Figure 2.3a provides a number of valuable insights intothe nature of intermolecular electrostatic interactions and the factors which influence them:

1. They may be either attractive or repulsive depending on the sign of the charges.If the charges have similar signs, wE(s) is positive and the interaction is repulsive,but if they have opposite signs, wE(s) is negative and the interaction is attractive.

2. Their strength depends on the magnitudes of the charges involved (Q1 and Q2).Thus, ion–ion interactions are stronger than ion–dipole interactions, which are inturn stronger than dipole–dipole interactions. In addition, the strength of interac-tions involving ions increases as their valency increases, whereas the strength ofinteractions involving polar species increases as their dipole moment increases.

TABLE 2.1Parameters Needed to Calculate the Interaction PairPotential for Ion–Ion, Ion–Dipole, and Dipole–DipoleElectrostatic Interactions Using Equation 2.1 (see alsoFigure 2.3a)

Interaction type Example Q1Q2 n

Ion–ion Na+ Cl– (z1e) (z2e) 1Ion–dipole Na+ H2O (z1e) µ2 cos φ 2Dipole–dipole H2O H2O µ1µ2 f (φ) 3

Note: z is the valence, µ is the dipole moment, e is the electronic charge, and φis the angle between the charges.


(a)

3. Their strength increases as the center-to-center separation of the charged speciesdecreases. Thus, interactions between small ions or molecules (which can get closetogether) are stronger than those between large ions or molecules of the samecharge.

4. The range of ion–ion (1/s) interactions is longer than that of ion–dipole interactions(1/s2), which is longer than that of dipole–dipole interactions (1/s3).

5. Their strength depends on the nature of the material separating the charges (via εR):the higher the relative dielectric constant, the weaker the interaction (Table 2.2).Electrostatic interactions between two charged species in water (εR = 80) are

(b)

FIGURE 2.3 Dependence of the intermolecular pair potential on intermolecular separation for (a)electrostatic, (b) van der Waals, and (c) steric overlap interactions.


therefore much weaker than those between the same species in oil (εR = 2), whichaccounts for the much higher solubility of salts in water than in nonpolar solvents(Israelachvili 1992).

6. Their strength depends on the orientation of any dipoles involved, being strongestwhen partial charges of opposite sign are brought close together. When the elec-trostatic interaction between a dipole and another charged species is much strongerthan the thermal energy (Section 2.5), the dipole becomes permanently aligned soas to maximize the strength of the attraction. This alignment of dipoles is respon-sible for the high degree of structural organization of molecules in bulk water andthe ordering of water molecules around ions in aqueous solutions (Chapter 4).

The ionization of many biological molecules depends on the pH of the surroundingaqueous phase, and so electrostatic interactions involving these molecules are particularlysensitive to pH (Nakamura 1996; Damodaran 1989, 1994, 1996, 1997). For example, aprotein may exist as an individual molecule in solution when the pH is sufficiently far fromits isoelectric point because of strong electrostatic repulsion between the protein molecules,but it may precipitate when the pH of the solution is close to its isoelectric point because theelectrostatic repulsion between the molecules is no longer strong enough to prevent themfrom aggregating (Kinsella 1982, Kinsella and Whitehead 1989, Damodaran 1996). Thestrength of electrostatic interactions between molecules suspended in an aqueous solution isalso sensitive to the type and concentration of electrolyte present (Bergethon and Simons1990, de Wit and van Kessel 1996). A charged molecule tends to be surrounded by oppositelycharged ions (counterions), which effectively “screen” (reduce) the electrostatic interactionbetween other molecules of the same type (Chapter 3).

Electrostatic interactions play an extremely important role in determining the overallproperties of food emulsions because many of the major constituents of these products areeither ionic or dipolar (e.g., water, sugars, salts, proteins, polysaccharides, surfactants, acids,and bases) (Fennema 1996a). The unique physiochemical properties of water are governed

(c)

FIGURE 2.3 (continued)


TABLE 2.2Compilation of Molecular Properties of Some Common Liquidsand Solutes Needed to Calculate Intermolecular Interactions

Static Relative Dielectric Constants (R)

Water 78.5 Chloroform 4.8Ethylene glycol 40.7 Edible oils 2.5Methanol 32.6 Carbon tetrachloride 2.2Ethanol 24.3 Liquid paraffin 2.2Acetone 20.7 Dodecane 2.0Propanol 20.2 Hexane 1.9Acetic acid 6.2 Ai r 1.0

Molecular Diameters, Polarizabilities, and Dipole Moments

Molecule type (nm) /40 (×××××10–30 m3) (D a)

H2O 0.28 1.48 1.85CH4 0.40 2.60 0HCl 0.36 2.63 1.08CH3Cl 0.43 4.56 1.87CCl4 0.55 10.5 0NH3 0.36 2.26 1.47Methanol 0.42 3.2 1.69Ethanol b 5.2 1.69Acetone b 6.4 2.85Benzene 0.53 10.4 0

a D = 3.336 × 10–30 C m.

b Cannot be treated as spheres.

Taken from Israelachvili 1992 and Buffler 1995.

by relatively strong dipole–dipole interactions which cause the water molecules to becomehighly organized (Chapter 4). The accumulation and organization of water molecules aroundsolutes are determined by various types of dipole–dipole, dipole–ion, and ion–ion interactions(Chapter 4). The “screening” of electrostatic interactions between charged emulsion dropletsis due to the attraction of counterions to the surface of the droplets (Chapters 3 and 7). Theconformation and interactions of biopolymers in aqueous solution are governed by electro-static interactions between the charged groups and the surrounding molecules (Chapter 4).These examples highlight the importance of understanding the origin and nature of electro-static interactions in food emulsions.

2.3.3. van der Waals Interactions

van der Waals forces act between all types of molecular species, whether they are ionic, polar,or nonpolar (Hiemenz 1986, Israelachvili 1992). They are conveniently divided into threeseparate contributions, which all rely on the polarization of molecules (Figure 2.4):

1. Dispersion forces. These forces arise from the interaction between an instanta-neous dipole and a dipole induced in a neighboring molecule by the presence ofthe instantaneous dipole. The electrons in a molecule are continually movingaround the nucleus. At any given instant in time, there is an uneven distribution of


the negatively charged electrons around the positively charged nucleus, and so aninstantaneous dipole is formed. This instantaneous dipole generates an electricalfield which induces a dipole in a neighboring molecule. Consequently, there is aninstantaneous attractive force between the two dipoles. On average, the attractionbetween the molecules is therefore finite, even though the average net charge onthe molecules involved is zero.

2. Induction forces. These forces arise from the interaction between a permanentdipole and a dipole induced in a neighboring molecule by the presence of thepermanent dipole. A permanent dipole causes an alteration in the distribution ofelectrons of a neighboring molecule, which leads to the formation of an induceddipole. The interaction between the permanent dipole and the induced dipole leadsto an attractive force between the molecules.

3. Orientation forces. These forces arise from the interaction between two perma-nent dipoles that are continuously rotating. On average, these rotating dipoles haveno net charge, but there is still a weak attractive force between them because themovement of one dipole induces some correlation in the movement of a neighbor-ing dipole. When the interaction between the two dipoles is strong enough to causethem to be permanently aligned, this contribution is replaced by the electrostaticdipole–dipole interaction described in the previous section.

As will be seen in the next chapter, an understanding of the origin of these three contri-butions to the van der Waals interaction has important consequences for predicting thestability of emulsion droplets to aggregation.

The overall intermolecular pair potential due to van der Waals interactions is given by:

w sC C C

sRVDW

disp ind orient

0(4( )

( )

)=

− + +πε ε 2 6 (2.2)

where Cdisp, Cind, and Corient are constants which depend on the dispersion, induction, andorientation contributions, respectively (Hiemenz 1986). Their magnitude depends on thedipole moment (for permanent dipoles) and the polarizability (for induced dipoles) of themolecules involved in the interaction (Table 2.2). The polarizability is a measure of thestrength of the dipole induced in a molecule when it is in the presence of an electrical field:the larger the polarizability, the easier it is to induce a dipole in a molecule. For mostbiological molecules, the dominant contribution to the van der Waals interaction is the

FIGURE 2.4 Schematic representation of van der Waals intermolecular interactions which involveeither the electronic or orientational polarization of molecules.


dispersion force, with the important exception of water, where the major contribution is fromthe orientation force (Israelachvili 1992). Examination of Equation 2.2 and Figure 2.3bprovides some useful physical insights into the factors that influence the van der Waalsinteractions between molecules:

1. The value of Cdisp, Cind, and Corient is always positive, which means that the overallinteraction potential between two molecules is always negative (attractive).

2. The interaction is relatively short range, decreasing rapidly with intermolecularseparation (1/s6).

3. The attraction between molecules decreases as the dielectric constant of the inter-vening medium increases, which highlights the electromagnetic origin of van derWaals interactions.

4. The magnitude of the interaction increases as the polarizability and dipole momentof the molecules involved increase.

Although van der Waals interactions act between all types of molecular species, they areconsiderably weaker than electrostatic interactions (Figure 2.3 and Table 2.3). For thisreason, they are most important in determining interactions between nonpolar molecules,where electrostatic interactions do not make a significant contribution. Indeed, the structureand physicochemical properties of organic liquids are largely governed by the van der Waalsinteractions between the molecules (Israelachvili 1992).

All types of van der Waals interaction involve either the electronic or orientational polar-ization of molecules and have a 1/s6 dependence on intermolecular separation (Hiemenz1986). Another type of interaction that depends on molecular polarization but which does nothave a 1/s6 dependence on intermolecular separation is ion polarization (Israelachvili 1992).Although this type of interaction is not strictly a van der Waals interaction, it is convenientto consider it in this section because it also involves polarization. A positively charged ioncauses the electrons in a neighboring molecule to be pulled toward it, thus inducing a dipolewhose δ– pole faces toward the ion. Similarly, a negatively charged ion causes the electronsin a neighboring molecule to be repelled away from it, thus inducing a dipole whose δ+ polefaces toward the ion. Thus there is an attractive force between the ion and the induced dipolebecause of electronic polarization. For polar molecules, there may be an additional contribu-tion due to orientational polarization. When the interaction between an ion and a dipole is notstrong enough to cause the dipole to become permanently aligned, the dipole continuouslyrotates because of its thermal energy (Section 2.5). On average, there is no net charge on arotating dipole because of its continuous rotation, but in the presence of an ion there is a netattraction between the ion and the dipole because the low-energy orientations are preferred(Israelachvili 1992). When the interaction between the ion and the dipole is strong enoughto cause the dipole to be permanently aligned, this contribution should be replaced by theelectrostatic ion–dipole interaction described in the previous section.

The intermolecular pair potential for ion polarization is given by:

w sze

s kTIPR

( )( )

( )= − +

2

02 4 0

2

2 4 3πε εα µ

(2.3)

where the α0 term is the contribution from the electronic polarizability of the molecule andthe µ2/3kT term is the contribution from the orientational polarizability. For nonpolar mol-ecules, only the electronic polarization term contributes to this interaction, but for polar


molecules, both electronic and orientational polarization contribute. This type of interactionis significantly stronger than the van der Waals interactions mentioned above and shouldtherefore be included in any calculation of the interaction energy between molecules involv-ing ions.

2.3.4. Steric Overlap Interactions

When two atoms or molecules come so close together that their electron clouds overlap, thereis an extremely large repulsive force generated between them (Figure 2.3c). This stericoverlap force is very short range and increases rapidly when the separation between the twomolecules becomes less than the sum of their radii (σ = r1 + r2). A number of empiricalequations have been derived to describe the dependence of the steric overlap intermolecularpair potential, wsteric(s), on molecular separation (Israelachvili 1992). The “hard-shell” modelassumes that the repulsive interaction is zero when the separation is greater than σ butinfinitely large when it is less than σ:

w sssteric( ) =

∞σ(2.4)

In reality, molecules are slightly compressible, and so the increase in the steric overlaprepulsion is not as dramatic as indicated by Equation 2.4. The slight compressibility ofmolecules is accounted for by a “soft-shell” model, such as the power-law model:

w sssteric( ) =

σ 12

(2.5)

At separations greater than σ, the steric overlap repulsion is negligible, but at separationsless than this value, there is a steep increase in the interaction pair potential, which means thatthe molecules strongly repel one another. The strong repulsion that arises from steric overlapdetermines the effective size of atoms and molecules and how closely they can come together.It therefore has a strong influence on the packing of molecules in liquids and solids.

2.4. OVERALL INTERMOLECULAR PAIR POTENTIAL

We are now in a position to calculate the overall interaction between a pair of molecules.Assuming that no chemical reactions occur between the molecules, the overall intermolecularpair potential is the sum of the various physical interactions mentioned above:

w s w s w s w sE( ) ( ) ( ) ( )= + +VDV steric (2.6)

The magnitude of each of the individual contributions to the overall interaction potentialis strongest at close separations and decreases as the molecules move apart. Nevertheless, theoverall intermolecular pair potential has a more complex dependence on separation, whichmay be attractive at some separations and repulsive at others, because it is the sum of anumber of interactions which each have different magnitudes, ranges, and signs.

To highlight some of the most important features of intermolecular interactions, it isuseful to consider the interaction of a pair of spherical nonpolar molecules (i.e., no electro-static interactions). The overall intermolecular pair potential for this type of system is given


by an expression known as the Lennard–Jones potential (Bergethon and Simons 1990,Baianu 1992):

w sA

sB

s( ) = − +

6 12 (2.7)

where the A term represents the contribution from the van der Waals interactions (Equation2.2) and the B term represents the contribution from the steric overlap interaction (Equation2.5). The dependence of the intermolecular pair potential on separation is illustrated in Figure2.5. The van der Waals interactions are attractive at all separations, whereas the steric overlapinteractions are repulsive. At large separations, w(s) is so small that there is no effectiveinteraction between the molecules. As the molecules are brought closer together, the pairpotential becomes increasingly attractive (negative) because the van der Waals interactionsdominate. Eventually, the molecules get so close together that their electron clouds overlapand the pair potential becomes strongly repulsive (positive) because steric overlap interac-tions dominate. Consequently, there is a minimum in the overall intermolecular pair potentialat some intermediate separation (s*). Two molecules will tend to remain associated in thispotential energy minimum in the absence of any disruptive influences (such as thermal energyor applied external forces), with a “bond length” of s* and a “bond strength” of w(s*).

It is often more convenient to describe the interaction between a pair of molecules in termsof forces rather than potential energies (Israelachvili 1992). The force acting between twomolecules can simply be calculated from the intermolecular pair potential using the followingrelationship: F(s) = –dw(s)/ds. The minimum in the potential energy curve therefore occursat a separation where the net force acting between the molecules is zero (i.e., the attractiveand repulsive forces exactly balance). If the molecules move closer together, they experiencea repulsive force, and if they move farther apart, they experience an attractive force.

FIGURE 2.5 Intermolecular pair potential for a pair of spherical nonpolar molecules. The curves werecalculated assuming typical values for the constants: A = 10–77 J m6 and B = 10–134 J m12.


2.5. BOND STRENGTHS AND THE ROLE OF THERMAL ENERGY

The molecules in a substance are in continual motion (translational, rotational, and vibra-tional) because of their thermal energy (kT) (Israelachvili 1992, Atkins 1994). The thermallyinduced movement of molecules has a disorganizing influence, which opposes the formationof intermolecular bonds. For this reason, the strength of intermolecular interactions is usuallyjudged relative to the thermal energy: kT ≈ 4.1 × 10–24 kJ per bond or RT ≈ 2.5 kJ mol–1. Ifthe bond strength is sufficiently greater than kT, the molecules will remain together, but if itis sufficiently smaller, they will tend to move apart. At intermediate bond strengths, themolecules spend part of their time together and part of their time apart (i.e., bonds are rapidlybreaking and reforming).

The bond strengths of a number of important types of intermolecular interaction aresummarized in Table 2.3. In a vacuum, the strength of these bonds decreases in the followingorder: ion–ion, covalent > ion–dipole > dipole–dipole > van der Waals. With the exception

TABLE 2.3Approximate Bond Strengths for Some of the Most Important Typesof Molecular Interactions That Occur in Foods at Room Temperature

In vacuum In water

Type of w(s*) w(s*) w(s*) w(s*)interaction (kJ mol–1) (RT ) (kJ mol–1) (RT )

Covalent bondsC–O 340 140C–C 360 140C–H 430 170O–H 460 180C=C 600 240C≡N 870 350Electrostatic ion–ionNa+ Cl– 500 200 6.3 2.5Mg2+ Cl– 1100 460 14.1 5.7Al3+ Cl– 1800 730 22.5 9.1Ion–dipoleNa+ H2O 97 39 1.2 0.5Mg2+ H2O 255 103 3.2 1.3Al3+ H2O 445 180 5.6 2.3Dipole–dipoleH2O H2O 38 15 0.5 0.2Ion polarizationNa+ CH4 24 10van der WaalsCH4 CH4 1.5 0.60C6H14 C6H14 7.4 3.0C12H26 C12H26 14.3 5.7C18H38 C18H38 21.2 6.1CH4 H2O 2.6 0.7H2O H2O 17.3 6.9

Note: All dipole interactions assuming that the molecules are aligned so they get maximumattraction. van der Waals forces calculated from Israelachvili (1992) assuming that w(s*) isapproximately equal to the cohesive energy over 6.


of methane (a small nonpolar molecule), the bonds between the molecules shown in Table2.3 are sufficiently strong (compared to the thermal energy) to hold them together in a liquidor solid. It must be stressed that the strength of the electrostatic and van der Waals interac-tions between molecules decreases appreciably when they are surrounded by a solvent ratherthan a vacuum, especially when the solvent has a high dielectric constant. The strength of theelectrostatic interaction between molecules dispersed in water is about 80 times less than thatin a vacuum because of the high relative dielectric constant of water (εR ≈ 80). This largelyaccounts for the high water solubility of many ionic crystals (Israelachvili 1992). The strengthof ion–dipole interactions between molecules dispersed in water is usually sufficiently large(compared to the thermal energy) to cause water molecules in the immediate vicinity of anion to be attracted to its surface and to become aligned, especially when the ion is small andhighly charged (Chapter 4). Even some types of dipole–dipole interaction are sufficientlystrong in water to cause a high degree of structural organization of the molecules (e.g., thetetrahedral structure formed in bulk water) (Chapter 4).

The strength of van der Waals interactions is also reduced when the molecules involvedare surrounded by a solvent (Israelachvili 1992). At large separations between the molecules,the van der Waals interaction between oil molecules is reduced by a factor of about 20 whenthey are surrounded by water rather than a vacuum (Israelachvili 1992). At present, there isno theory that can accurately account for the reduction in van der Waals interactions betweenmolecules which are in close contact within a solvent. Nevertheless, we can reasonablypostulate that van der Waals bonds in a solvent are very weak and that only for relativelylarge molecules will these interactions be strong enough to hold the molecules together. Thisaccounts for the fact that the volatility of molecules increases as their molecular weight orpolarity decreases (Israelachvili 1992).

2.6. THE STRUCTURAL ORGANIZATION OF MOLECULESIN LIQUIDS

2.6.1. Thermodynamics of Mixing

In food emulsions, we are usually concerned with the interactions of large numbers ofmolecules in a liquid, rather than between a pair of isolated molecules in a vacuum. We musttherefore consider the interaction of a molecule with its neighbors and how these interactionsdetermine the overall organization of the molecules within a liquid (Murrell and Boucher1982, Murrell and Jenkins 1994, Evans and Wennerstrom 1994). The behavior of largenumbers of molecules at equilibrium can be described by statistical thermodynamics (Searsand Salinger 1975, Atkins 1994). A molecular ensemble tends to organize itself so that themolecules are in an arrangement which minimizes the free energy of the system. The freeenergy of a molecular ensemble is governed by both enthalpy and entropy contributions(Bergethon and Simons 1990). The enthalpy contributions are determined by the molecularinteraction energies discussed above, while the entropy contributions are determined by thetendency of a system to adopt its most disordered state.

Consider a hypothetical system that consists of a collection of two different types ofequally sized spherical molecules, A and B (Figure 2.6). The free energy change that occurswhen these molecules are mixed is given by:

∆ ∆ ∆G E T Smix mix mix= − (2.8)

where ∆Emix and ∆Smix are the differences in the molecular interaction energy and entropy ofthe mixed and unmixed states, respectively. Practically, we may be interested in whether the


resulting system consists of two immiscible liquids or is a mixture where the molecules aremore or less intermingled (Figure 2.6). Thermodynamics tells us that if ∆Gmix is positive,mixing is unfavorable and the molecules tend to exist as two separate phases (i.e., they areimmiscible); if ∆Gmix is negative, mixing is favorable and the molecules tend to be inter-mingled with each other (i.e., they are miscible); and if ∆Gmix ≈ 0, the molecules are partlymiscible and partly immiscible. For simplicity, we assume that if the two types of moleculesdo intermingle with each other, they form a regular solution (i.e., a completely randomarrangement of the molecules) (Figure 2.6 right) rather than an ordered solution, in which thetype A molecules are preferentially surrounded by type B molecules or vice versa. In practice,this means that the attractive forces between the two different types of molecules are notmuch stronger than the thermal energy of the system (Atkins 1994, Evans and Wennerstrom1994). This argument is therefore only applicable to mixtures that contain nonpolar or slightlypolar molecules, where strong ion–ion or ion–dipole interactions do not occur. Despite thesimplicity of this model system, we can still gain considerable insight into the behavior ofmore complex systems that are relevant to food emulsions. In the following sections, weseparately consider the contributions of the interaction energy and the entropy to the overallfree energy change that occurs on mixing.

2.6.2. Potential Energy Change on Mixing

An expression for ∆Emix can be derived by calculating the total interaction energy of themolecules before and after mixing (Israelachvili 1992, Evans and Wennerstrom 1994). Forboth the mixed and the unmixed system, the total interaction energy is determined bysumming the contribution of each of the different types of bond:

E n w n w n w= + +AA AA BB BB AB AB (2.9)

where nAA, nBB, and nAB are the total number of bonds, and wAA, wBB, and wAB are theintermolecular pair potentials at equilibrium separation that correspond to interactions be-tween A–A, B–B, and A–B molecules, respectively. The total number of each type of bondformed is calculated from the number of molecules present in the system, the coordinationnumber of the individual molecules (i.e., the number of molecules in direct contact withthem), and their spatial arrangement. For example, many of the A–A and B–B interactionsthat occur in the unmixed system are replaced by A–B interactions in the mixed system. Thedifference in the total interaction energy between the mixed and unmixed states is thencalculated: ∆Emix = Emix – Eunmixed. This type of analysis leads to the following equation(Evans and Wennerstrom 1994).

∆E nX X wmix A B= (2.10)

FIGURE 2.6 System in which two types of molecules may be completely miscible or form a regularsolution depending on the strength of the interactions between them and the entropy of mixing.


where n is the total number of moles, w is the effective interaction parameter, and XA andXB are the mole fractions of molecules of type A and B, respectively. The effective interactionparameter is a measure of the compatibility of the molecules in a mixture and is related tothe intermolecular pair potential between isolated molecules by the expression

w zN w w w= − +A AB AA BB[ )](12 (2.11)

where z is the coordination number of a molecule and NA is Avogadro’s number. Theeffective interaction parameter determines whether the transfer of a molecule from a liquidwhere it is surrounded by similar molecules to one in which it is partly surrounded bydissimilar molecules is favorable (w is negative), unfavorable (w is positive), or indifferent(w = 0). It should be stressed that even though there may be attractive forces between all themolecules involved (i.e., wAA, wBB, and wAB may all be negative), the overall interactionpotential can be either negative (favorable to mixing) or positive (unfavorable to mixing)depending on the relative magnitude of the interactions. If the strength of the interactionbetween two different types of molecules (wAB) is greater (more negative) than the averagestrength between similar molecules (wAB < [wAA + wBB]/2), then w is negative, which favorsthe intermingling of the different types of molecules. On the other hand, if the strength of theinteraction between two different types of molecules is weaker (less negative) than theaverage strength between similar molecules (wAB > [wAA + wBB]/2), then w is positive, whichfavors phase separation. If the strength of the interaction between different types of moleculesis the same as the average strength between similar molecules (wAB = [wAA + wBB]/2), thenthe system has no preference for any particular arrangement of the molecules within thesystem. In summary, the change in the overall interaction energy may either favor or opposemixing, depending on the relative magnitudes of the intermolecular pair potentials.

2.6.3. Entropy Change on Mixing

An expression for ∆Smix is obtained from simple statistical considerations (Israelachvili 1992,Evans and Wennerstrom 1994). The entropy of a system depends on the number of differentways the molecules can be arranged. For an immiscible system, there is only one possiblearrangement of the two different types of molecules (i.e., zero entropy), but for a regularsolution, there are a huge number of different possible arrangements (i.e., high entropy). Astatistical analysis of this situation leads to the derivation of the following equation for theentropy of mixing:

∆S nR X X X Xmix A A B B= − +( ln ln ) (2.12)

∆Smix is always positive because XA and XB are both between zero and one (so that the naturallogarithm terms are negative), which reflects the fact that there is always an increase inentropy after mixing. For regular solutions, the entropy contribution (–T∆Smix) always de-creases the free energy of mixing (i.e., favors the intermingling of the molecules). It shouldbe stressed that for more complex systems, there may be additional contributions to theentropy due to the presence of some order within the mixed state (e.g., organization of watermolecules around a solute molecule) (Chapter 4).

2.6.4. Free Energy Change on Mixing

For a regular solution, the free energy change on mixing depends on the combined contribu-tions of the interaction energies and the entropy:


∆G n X X w RTX X X Xmix A B A A B B= + +[ ( ln ln )] (2.13)

We are now in a position to investigate the relationship between the strength of theinteractions between molecules and their structural organization. The dependence of the freeenergy of mixing on the effective interaction parameter and the composition of a systemconsisting of two different types of molecules is illustrated in Figure 2.7. The molecules arecompletely miscible when the free energy of mixing is negative and large compared to thethermal energy, are partly miscible when ∆Gmix ≈ 0, and are completely immiscible when thefree energy of mixing is positive and large compared to the thermal energy. Figure 2.7indicates that mixing occurs even when the effective interaction parameter is zero, becauseof the contribution of the entropy of mixing term. This accounts for the miscibility of liquidsin which the interactions between the two types of molecules are fairly similar (e.g., twononpolar oils). Two liquids are completely immiscible when the effective interaction param-eter is large and positive. The above approach enables us to use thermodynamic consider-ations to relate bulk physicochemical properties of liquids (such as immiscibility) to molecu-lar properties (such as the effective interaction parameter and the coordination number).

2.6.5. The Properties of More Complex Systems

The derivation of Equation 2.13 depends on making a number of simplifying assumptionsabout the properties of the system that are not normally valid in practice (e.g., that themolecules are spherical, that they all have the same size and coordination number, and thatthere is no ordering of the molecules within the mixture) (Israelachvili 1992). It is possible

FIGURE 2.7 Dependence of the free energy of mixing (calculated using Equation 2.13) on thecomposition and effective interaction parameter of a binary liquid. When ∆Gmix is much less than –RT,the system tends to be mixed; otherwise, it will be partly or wholly immiscible.


to incorporate some of these features into the above theory, but a more elaborate mathemati-cal analysis is required. Food molecules come in all sorts of different sizes, shapes, andflexibilities. They may be nonpolar, polar, or amphiphilic; they may have specific bindingsites; or they may have to be in a certain orientation before they can interact with theirneighbors. In addition, a considerable degree of structural organization of the moleculeswithin a solvent often occurs when a solute is introduced. The variety of molecular charac-teristics exhibited by food molecules accounts for the great diversity of structures that areformed in food emulsions, such as bulk liquids, regular solutions, organized solutions,micelles, molecular networks, and immiscible liquids (Figure 2.1).

Another problem with the thermodynamic approach is that food systems are rarely atthermodynamic equilibrium because of the presence of various kinetic energy barriers thatprevent the system from reaching its lowest energy state. This approach cannot therefore tellus whether or not two liquids will exist as an emulsion, because an emulsion is a thermody-namically unstable system. Nevertheless, it can tell us whether two liquids are capable offorming an emulsion (i.e., whether they are immiscible or miscible). Despite the obviouslimitations of the simple thermodynamic approach, it does highlight some of the mostimportant features of molecular organization, especially the importance of considering bothinteraction energies and entropy effects.

2.7. MOLECULAR INTERACTIONS AND CONFORMATION

So far, we have only considered the way that molecular interactions influence the spatialdistribution of molecules in a system. Molecular interactions can also determine the three-dimensional conformation and flexibility of individual molecules (Lehninger et al. 1993,Atkins 1994, Gelin 1994). Small molecules, such as H2O and CH4, normally exist in a singleconformation which is determined by the relatively strong covalent bonds that hold the atomstogether (Karplus and Porter 1970, Atkins 1994). On the other hand, many larger moleculescan exist in a number of different conformations because of the possibility of rotation aroundsaturated covalent bonds (e.g., proteins and polysaccharides) (Baianu 1992, Bergethon andSimons 1990, Lehninger et al. 1993, Fennema 1996a). A macromolecule will tend to adoptthe conformation that has the lowest free energy under the prevailing environmental condi-tions (Alber 1989). The conformational free energy of a molecule is determined by theinteraction energies and entropy of the system that contains it (Dill 1990). The molecularinteractions may be between different parts of the same molecule (intramolecular) or betweenthe molecule and its neighbors (intermolecular). Similarly, the entropy is determined by thenumber of conformations that the molecule can adopt, as well as by any changes in theentropy caused by interactions with its neighbors (e.g., restriction of their translational orrotational motion) (Alber 1989, Dill 1990).

To highlight the importance of molecular interactions and entropy in determining theconformation of molecules in solution, it is useful to examine a specific example. Considera hydrophilic biopolymer molecule in an aqueous solution that can exist in either a helicalor a random-coil conformation depending on the environmental conditions (Figure 2.8).Many types of food biopolymers are capable of undergoing this type of transformation,including the protein gelatin (Walstra 1996b) and the polysaccharide xanthan (BeMiller andWhistler 1996). The free energy associated with the transition (helix ⇔ coil) between thesetwo different conformations is given by:

∆ ∆ ∆G E T Sh c h c h c→ → →= − (2.14)


where ∆Gh→c, ∆Eh→c, and ∆Sh→c are the free energy, interaction energy, and entropy changesassociated with the helix-to-coil transformation. If ∆Gh→c is negative, the random-coilconformation is favored; if ∆Gh→c is positive, the helix conformation is favored; and if∆Gh→c ≈ 0, the molecule spends part of its time in each of the conformations. A helicalconformation often allows a molecule to maximize the number of energetically favorableintermolecular and intramolecular interactions while minimizing the number of energeti-cally unfavorable ones (Bergethon and Simons 1990, Dickinson and McClements 1995).Nevertheless, it has a much lower entropy than the random-coil state because the moleculecan only exist in a single conformation, whereas in the random-coil state the molecule canexist in a large number of different conformations that have similar low energies. At lowtemperatures, the interaction energy term dominates the entropy term and so the moleculetends to exist as a helix, but as the temperature is raised, the entropy term (–T∆Sh→c)becomes increasingly important until eventually it dominates and the molecule unfolds. Thetemperature at which the helix-to-coil transformation takes place is referred to as thetransition temperature (Th→c), which occurs when ∆Gh→c = 0. Similar arguments can be usedto account for the unfolding of globular proteins when they are heated above a particulartemperature, although the relative contribution of the various types of interaction energy isdifferent (Dickinson and McClements 1995). It must be stressed that many food moleculesare unable to adopt their thermodynamically most stable conformation because of thepresence of various kinetic energy barriers (Section 1.2.1). When an energy barrier is muchgreater than the thermal energy of the system, a molecule may be “trapped” in a metastablestate indefinitely.

The flexibility of molecules in solution is also governed by both thermodynamic andkinetic factors. Thermodynamically, a flexible molecule must be able to exist in a number ofconformations that have fairly similar (±kT) low free energies. Kinetically, the energybarriers that separate these energy states must be small compared to the thermal energy of thesystem. When both of these criteria are met, a molecule will rapidly move between a numberof different configurations and therefore be highly flexible. If the free energy differencebetween the conformations is large compared to the thermal energy, the molecule will tendto exist predominantly in the minimum free energy state (unless it is locked into a metastablestate by the presence of a large kinetic energy barrier).

Knowledge of the conformation and flexibility of a macromolecule under a particular setof environmental conditions is particularly important in understanding and predicting thebehavior of many ingredients in food emulsions. The conformation and flexibility of amolecule determine its chemical reactivity, catalytic activity, intermolecular interactions, and

FIGURE 2.8 The conformation of a molecule in solution is governed by a balance of interactionenergies and entropic effects. A helical molecule unfolds when it is heated above a certain temperaturebecause the random-coil conformation is entropically more favorable than the helical conformation.


functional properties (e.g., solubility, dispersability, water-holding capacity, gelation, foam-ing, and emulsification) (Damodaran 1994, 1996, 1997).

2.8. HIGHER ORDER INTERACTIONS

When one consults the literature dealing with molecular interactions in foods and otherbiological systems, one often comes across the terms “hydrogen bonding” and “hydrophobicinteractions” (Bergethon and Simons 1990, Baianu 1992, Fennema 1996a). In reality, theseterms are a shorthand way of describing certain combinations of interactions which occurbetween specific chemical groups commonly found in food molecules. Both of these higherorder interactions consist of contributions from various types of interaction energy (van derWaals, electrostatic, and steric overlap), as well as some entropy effects. It is useful tohighlight the general features of hydrogen bonds and hydrophobic interactions in this section,before discussing their importance in determining the properties of individual food compo-nents later (Chapter 4).

2.8.1. Hydrogen Bonds

Hydrogen bonds play a crucial role in determining the functional properties of many of themost important molecules present in food emulsions, including water, proteins, lipids, carbo-hydrates, surfactants, and minerals (Chapter 4). They are formed between a lone pair ofelectrons on an electronegative atom (such as oxygen) and a hydrogen atom on a neighbor-ing group (i.e., O–Hδ+ … Oδ–) (Baker and Hubbard 1984, Baianu 1990, Bergethon andSimons 1990, Lehninger et al. 1993). The major contribution to hydrogen bonds is electro-static (dipole–dipole), but van der Waals forces and steric repulsion also make a significantcontribution (Dill 1990). Typically, they have bond strengths between about 10 and 40 kJmol–1 and lengths of about 0.18 nm (Israelachvili 1992). The actual strength of a particularhydrogen bond depends on the electronegativity and orientation of the donor and acceptorgroups (Baker and Hubbard 1984). Hydrogen bonds are stronger than most other examplesof dipole–dipole interaction because hydrogen atoms have a strong tendency to becomepositively polarized and because they have a small radius. In fact, hydrogen bonds are sostrong that they cause appreciable alignment of the molecules involved. The strength anddirectional character of hydrogen bonds are responsible for many of the unique properties ofwater (Chapter 4).

2.8.2. Hydrophobic Interactions

Hydrophobic interactions also play a major role in determining the behavior of many impor-tant ingredients in food emulsions, particularly lipids, surfactants, and proteins (Nakai and Li-Chan 1988). They manifest themselves as a strong attractive force that acts between nonpolargroups separated by water (Ben-Naim 1980, Tanford 1980, Israelachvili 1992). Nevertheless,the actual origin of hydrophobic interactions is the ability of water molecules to formrelatively strong hydrogen bonds with their nearest neighbors, whereas nonpolar moleculescan only form relatively weak van der Waals bonds (Israelachvili 1992). When a nonpolarmolecule is introduced into liquid water, it causes the water molecules in its immediatevicinity to rearrange themselves, which changes both the interaction energy and entropy ofthe system (Chapter 4). It turns out that these changes are thermodynamically unfavorable,and so the system attempts to minimize contact between water and nonpolar groups, whichappears as an attractive force between the nonpolar groups (Ben-Naim 1980, Evans and


Wennerstrom 1994). It is this effect that is largely responsible for the immiscibility of oil andwater, the adsorption of surfactant molecules to an interface, the aggregation of proteinmolecules, and the formation of surfactant micelles, and it is therefore particularly importantfor food scientists to have a good understanding of its origin and the factors which influenceit (Nakai and Li-Chan 1988).

2.9. COMPUTER MODELING OF LIQUID PROPERTIES

Our understanding of the way that molecules organize themselves in a liquid can be greatlyenhanced by the use of computer modeling techniques (Murrell and Jenkins 1994, Gelin1994). Computer simulations of molecular properties have provided a number of valuableinsights that are relevant to a better understanding of the behavior of food emulsions,including the miscibility/immiscibility of liquids, the formation of surfactant micelles, theadsorption of emulsifiers at an interface, the transport of nonpolar molecules through anaqueous phase, and the conformation and flexibility of biopolymers in solution (van Gunsteren1988; Brady 1989; Ludescher 1990; Kumosinski et al. 1991a,b; Dickinson and McClements1995; Esselink et al. 1994; Gelin 1994). The first step in a molecular simulation is to definethe characteristics of the molecules involved (size, shape, flexibility, and polarity) and thenature of the intermolecular pair potentials that act between them (Gelin 1994).* A collectionof these molecules is arbitrarily distributed within a “box” that represents a certain region ofspace, and the change in the structural organization of the molecules is then monitored as theyare allowed to interact with each other. Depending on the simulation technique used, one canobtain information about the evolution of the structure with time and/or about the equilibriumstructure of the molecular ensemble. The two most commonly used computer simulationtechniques are the Monte Carlo approach and the molecular dynamics approach (Murrell andBoucher 1982, Murrell and Jenkins 1994).

2.9.1. Monte Carlo Techniques

This technique is named after Monte Carlo, a town in southern France which is famous forits gambling and casinos. The reason for this peculiar name is the fact that the movement ofthe molecules in the “box” is largely determined by a random selection process, just as thewinner in a roulette game is selected. Initially, one starts with an arbitrary arrangement of themolecules in the box. The overall interaction energy is then calculated from a knowledge ofthe positions of all the molecules and their intermolecular pair potentials. One of the mol-ecules is then randomly selected and moved to a new location and the overall interactionenergy is recalculated. If the energy decreases, the move is definitely allowed, but if itincreases, the probability of the move being allowed depends on the magnitude of the energychange compared to the thermal energy. When the increase in energy is much greater thanRT, the move is highly unlikely and will probably be rejected, but if it is on the same orderas RT, it is much more likely to be accepted. This procedure is continued until there is nofurther change in the average energy of the system after moving a molecule, which is takento be the minimum potential energy state of the system. Entropy effects are accounted for bycontinuously monitoring the fluctuations in the overall interaction energy after successivemolecules have been moved once the system has reached equilibrium. At thermodynamicequilibrium, the probability of finding a molecular ensemble in a particular potential energy

* It should be noted that it is usually necessary to make a number of simplifying assumptions about the propertiesand interactions of the molecules in order to create computer programs which can be solved in a reasonable timeperiod.


level (E) is proportional to the Boltzmann factor (P ∝ e–E/RT). Thus the free energy of thesystem is found by calculating the average value of e–E/RT over all the possible states of thesystem:

G RT e E RT= − −ln / (2.15)

where < > represents the average over all possible states of the system. The Monte Carlotechnique therefore provides information about the equilibrium properties of a system, ratherthan about the evolution in its properties with time.

2.9.2. Molecular Dynamics Techniques

This technique is named for the fact that it relies on monitoring the movement of moleculeswith time. Initially, one starts with an arbitrary arrangement of the molecules in the box. Thecomputer then calculates the force which acts on each of the molecules as a result of itsinteractions with the surrounding molecules. Newton’s equations of motion are then used todetermine the direction and speed at which each of the molecules moves within a timeinterval that is short compared to the average time between molecular collisions (typically10–15 to 10–14 s). By carrying out the computation over a large number of successive timeintervals, it is possible to monitor the evolution of the system with time. At present, ittypically takes about 1 s of computer time to simulate 1 ps (10–12 s) of real time (Israelachvili1992). Events at the molecular level typically occur over time scales ranging from a fewnanoseconds (e.g., molecular rotation) to a few microseconds (e.g., collisions of colloidalparticles), which corresponds to computer times ranging from a few minutes to a few daysor weeks (Israelachvili 1992). As computer technology advances, these computation timeswill inevitably decrease, but they are still likely to be too long to allow computer modelingto be routinely used to study many important molecular processes. A molecular dynamicssimulation should lead to the same final state as a Monte Carlo simulation if it is allowed toproceed long enough to reach equilibrium. The free energy of the system is determined bythe same method as for Monte Carlo simulations (i.e., by taking into account the fraction ofmolecules that occupies each energy state once the system has reached equilibrium) (Equa-tion 2.15).

In practice, molecular dynamics simulations are more difficult to set up and take muchlonger to reach equilibrium than Monte Carlo simulations. For this reason, Monte Carlosimulations are more practical if a researcher is only interested in equilibrium properties, butmolecular dynamics simulations are used when information about both the kinetics andthermodynamics of a system is required. Molecular dynamics techniques are particularlysuitable for studying nonequilibrium processes, such as mass transport, fluid flow, adsorptionkinetics, and solubilization processes (Dickinson and McClements 1995). It is clear from theabove discussion that each technique has its own advantages and disadvantages and that bothtechniques can be used to provide useful insights into the molecular basis for the bulkphysiochemical properties of food emulsions.


3 Colloidal Interactions

3.1. INTRODUCTION

Food emulsions are microheterogeneous materials that contain a variety of different structuralentities which range in size, shape, and physicochemical properties, including atoms, mol-ecules, molecular aggregates, micelles, emulsion droplets, crystals, and air cells (Dickinsonand Stainsby 1982, Dickinson 1992). Many of these structural entities have at least onedimension that falls within the colloidal size range (i.e., between a few nanometers and a fewmicrometers) (see Figure 1.8). The characteristics of these colloidal particles, and theirinteractions with each other, are responsible for many of the most important physicochemicaland organoleptic properties of food emulsions. The ability of food scientists to understand,predict, and control the properties of food emulsions therefore depends on a knowledge of theinteractions that arise between colloidal particles. In this chapter, we examine the origin andnature of the most important types of colloidal interaction, while in later chapters we considerthe relationship between these interactions and the stability, rheology, and appearance of foodemulsions (Chapters 7 to 9). This is one of the most exciting and rewarding areas of researchcurrently being pursued in emulsion science, and our understanding is rapidly advancing asa result of recent developments in computer modeling, experimental techniques, and theconcerted endeavors of individuals working in a variety of scientific disciplines, includingmathematics, physics, chemistry, biology, and food science.

The interaction between a pair of colloidal particles is the result of interactions betweenall of the molecules within them, as well as those within the intervening medium (Hunter1986, Israelachvili 1992). For this reason, many of the interactions between colloidal particlesappear at first glance to be similar to those between molecules (e.g., van der Waals, electro-static, and steric) (Chapter 2). Nevertheless, the characteristics of these colloidal interactionsare often different from their molecular counterparts, because of additional features that arisedue to the relatively large size of colloidal particles compared to individual molecules. Themajor emphasis of this chapter will be on interactions between emulsion droplets, althoughthe same principles can be applied to the various other types of colloidal particles that arecommonly found in foods.

3.2. COLLOIDAL INTERACTIONS AND DROPLET AGGREGATION

Colloidal interactions govern whether emulsion droplets aggregate or remain as separateentities, as well as determine the characteristics of any aggregates formed (e.g., their size,shape, porosity, and deformability) (Dickinson 1992, Dickinson and McClements 1995,Bijsterbosch et al. 1995). Many of the bulk physicochemical and organoleptic properties offood emulsions are determined by the degree of droplet aggregation and the characteristicsof the aggregates (Chapters 7 to 9). It is therefore extremely important for food scientists tounderstand the relationship among colloidal interactions, droplet aggregation, and bulkproperties.


In Chapter 2, the interaction between two isolated molecules was described in terms of anintermolecular pair potential. In a similar fashion, the interactions between two emulsiondroplets can be described in terms of an interdroplet pair potential. The interdroplet pairpotential, w(h), is the energy required to bring two emulsion droplets from an infinite distanceapart to a surface-to-surface separation of h (Figure 3.1). Before examining specific types ofinteractions between emulsion droplets, it is useful to examine the features of colloidalinteractions in a more general fashion.

Consider a system which consists of two emulsion droplets of radius r at a surface-to-surface separation h (Figure 3.1). For convenience, we will assume that only two types ofinteractions occur between the droplets, one attractive and one repulsive:

w h w h w h( ) ( ) ( )= +attractive repulsive (3.1)

The overall interaction between the droplets depends on the relative magnitude and range ofthe attractive and repulsive interactions. A number of different types of behavior can bedistinguished depending on the nature of the interactions involved (Figure 3.2):

1. Attractive interactions dominate at all separations. If the attractive interactionsare greater than the repulsive interactions at all separations, then the overall inter-action is always attractive (Figure 3.2A), which means that the droplets will tendto aggregate (provided the strength of the interaction is greater than the disorga-nizing influence of the thermal energy).

2. Repulsive interactions dominate at all separations. If the repulsive interactionsare greater than the attractive interactions at all separations, then the overallinteraction is always repulsive (Figure 3.2B), which means that the droplets tendto remain as individual entities.

3. Attractive interactions dominate at large separations, but repulsive interactionsdominate at short separations. At very large droplet separations, there is noeffective interaction between the droplets. As the droplets move closer together,the attractive interaction initially dominates, but at closer separations the repulsiveinteraction dominates (Figure 3.2C). At some intermediate surface-to-surface sepa-ration, there is a minimum in the interdroplet interaction potential (hmin). The depthof this minimum, w(hmin), is a measure of the strength of the interaction betweenthe droplets, while the position of the minimum (hmin) corresponds to the mostlikely separation of the droplets. Droplets aggregate when the strength of theinteraction is large compared to the thermal energy, w(hmin) >> kT; remain asseparate entities when the strength of the interaction is much smaller than thethermal energy, w(hmin) << kT; and spend some time together and some time apartat intermediate interaction strengths, w(hmin) ≈ kT. When droplets fall into a deep

FIGURE 3.1 Emulsion droplets of radius r separated by a surface-to-surface separation h.


potential energy minimum, they are said to be strongly flocculated or coagulatedbecause a large amount of energy is required to pull them apart again. When theyfall into a shallow minimum, they are said to be weakly flocculated because theyare fairly easy to pull apart. The fact that there is an extremely large repulsionbetween the droplets at close separations prevents them from coming close enoughtogether to coalesce.

4. Repulsive interactions dominate at large separations, but attractive interactionsdominate at short separations. At very large droplet separations, there is noeffective interaction between the droplets. As the droplets move closer together,the repulsive interaction initially dominates, but at closer separations the attractiveinteraction dominates (Figure 3.2D). At some intermediate surface-to-surface sepa-ration (hmax), there is an energy barrier which the droplets must overcome beforethey can move any closer together. If the height of this energy barrier is largecompared to the thermal energy of the system, w(hmax) >> kT, the droplets areeffectively prevented from coming close together and will therefore remain asseparate entities. If the height of the energy barrier is small compared to thethermal energy, w(hmax) << kT, the droplets easily have enough thermal energy to“jump” over it, and they rapidly fall into the deep minimum that exists at closeseparations. At intermediate values, w(hmax) ≈ kT, the droplets still tend to aggre-gate, but this process occurs slowly because only a fraction of droplet–droplet

FIGURE 3.2 The interaction of a pair of emulsion droplets depends on the relative magnitude andrange of attractive and repulsive interactions.


collisions has sufficient energy to “jump” over the energy barrier. The fact thatthere is an extremely strong attraction between the droplets at close separations islikely to cause them to coalesce (i.e., merge together).

Despite the simplicity of the above model (Equation 3.1), we have already gained a numberof valuable insights into the role that colloidal interactions play in determining whetheremulsion droplets are likely to be unaggregated, flocculated, or coalesced. In particular, theimportance of the sign, magnitude, and range of the colloidal interactions has becomeapparent. As would be expected, the colloidal interactions that arise between the droplets inreal food emulsions are much more complex than those considered above (Dickinson 1992).First, there are a number of different types of repulsive and attractive interaction that contrib-ute to the overall interaction potential, each with a different sign, magnitude, and range.Second, food emulsions contain a huge number of droplets and other colloidal particles thathave different sizes, shapes, and properties. Third, the liquid that surrounds the droplets maybe compositionally complex, containing various types of ions and molecules. Droplet–dropletinteractions in real food emulsions are therefore influenced by the presence of the neighbor-ing droplets, as well as by the precise nature of the surrounding liquid. For these reasons, itis difficult to accurately account for colloidal interactions in real food emulsions because ofthe mathematical complexity of describing interactions between huge numbers of molecules,ions, and particles (Dickinson 1992). Nevertheless, considerable insight into the factorswhich determine the properties of food emulsions can be obtained by examining the interac-tion between a pair of droplets. In addition, our progress toward understanding complex foodsystems depends on first understanding the properties of simpler model systems. These modelsystems can then be incrementally increased in complexity and accuracy as advances aremade in our knowledge.

In the following sections, the origin and nature of the major types of colloidal interactionwhich arise between emulsion droplets are reviewed. In Section 3.11, we then consider waysin which these individual interactions combine with each other to determine the overallinterdroplet pair potential and thus the stability of emulsion droplets to aggregation. Aknowledge of the contribution that each of the individual colloidal interactions makes to theoverall interaction enables one to identify the most effective means of controlling the stabilityof a given system to aggregation.

3.3. VAN DER WAALS INTERACTIONS

3.3.1. Origin of van der Waals Interactions

Intermolecular van der Waals interactions arise because of the attraction between moleculesthat have been electronically or orientationally polarized (Section 2.5). In addition to actingbetween individual molecules, van der Waals interactions also act between macroscopicbodies that contain large numbers of molecules, such as emulsion droplets (Hiemenz 1986).The van der Waals interactions between macroscopic bodies can be calculated using twodifferent mathematical approaches (Hunter 1986, Derjaguin et al. 1987, Israelachvili 1992).In the microscopic approach, the van der Waals interaction between a pair of droplets iscalculated by carrying out a pairwise summation of the interaction energies of all the mol-ecules in one of the droplets with all of the molecules in the other droplet. Calculations madeusing this approach rely on a knowledge of the properties of the individual molecules, suchas polarizabilities, dipole moments, and electronic energy levels. In the macroscopic ap-proach, the droplets and surrounding medium are treated as continuous liquids which interactwith each other because of the fluctuating electromagnetic fields generated by the movement


of the electrons within them. Calculations made using this approach rely on a knowledge ofthe bulk physicochemical properties of the liquids, such as dielectric constants, refractiveindices, and absorption frequencies. Under certain circumstances, both theoretical approachesgive similar predictions of the van der Waals interaction between macroscopic bodies. Ingeneral, however, the macroscopic approach is usually the most suitable for describinginteractions between emulsion droplets because it automatically takes into account the effectsof retardation and the liquid surrounding the droplets (Hunter 1986).

3.3.2. Interdroplet Pair Potential

The van der Waals interdroplet pair potential, wVDW(h), of two emulsion droplets of equalradius r separated by a surface-to-surface distance h is given by the following expression(Figure 3.1):

w hA r

h rhr

h rh rh rh

h rh rVDW ( ) ln=−

+

+

+ +

+ +

+ +

1212

2

2

2 2

2

2 262

424 4

44 4

(3.2)

where A121 is the Hamaker function for emulsion droplets (medium 1) separated by a liquid(medium 2). The value of the Hamaker function can be calculated using either the micro-scopic or macroscopic approach mentioned above (Mahanty and Ninham 1976). At closeseparations (h << r), the above equation can be simplified considerably:

w hA r

hVDW ( ) =− 121

12(3.3)

This equation indicates that van der Waals interactions between colloidal particles (w ∝1/h) are much longer range than those between molecules (w ∝ 1/s6), which has importantconsequences for determining the stability of food emulsions. Equations for calculating thevan der Waals interaction between spheres of unequal radius are given by Hiemenz (1986).

3.3.3. Hamaker Function

In general, an accurate calculation of the Hamaker function of a pair of emulsion droplets isa complicated task (Mahanty and Ninham 1976, Hunter 1986, Israelachvili 1992). A knowl-edge of the optical properties (dielectric permittivities) of the oil, water, and interfacial phasesover a wide range of frequencies is required, and this information is not readily available formost substances (Hunter 1986, Roth and Lenhoff 1996). In addition, the full theory must besolved numerically using a digital computer (Pailthorpe and Russel 1982). Nevertheless,approximate expressions for the Hamaker function have been derived, which can be calcu-lated using data that can easily be found in the literature (Israelachvili 1992):

A A Av v121 0 0= += > (3.4)

where

A kTs

Ahv n n

n nv

s

s

ve

==

∞=

−+

=−( )

+( )∑0 31

1 2

1 2

2

012

22 2

12

22 3 2

34

1 3

16 2

ε εε ε > /


Here, ε is the static relative dielectric constant, n is the refractive index, ve is the majorelectronic absorption frequency in the ultraviolet region of the electromagnetic spectrum(which is assumed to be equal for both phases), h is Planck’s constant, and the subscripts 1and 2 refer to the continuous phase and droplets, respectively. Equation 3.4 indicates that theHamaker function of two similar droplets is always positive, which means that wVDW(h) isalways negative, so that the van der Waals interaction is always attractive. It should be noted,however, that the interaction between two colloidal particles containing different materialsmay be either attractive or repulsive, depending on the relative physical properties of theparticles and intervening medium (Israelachvili 1992, Milling et al. 1996).

In Equation 3.4, the Hamaker function is divided into two contributions: a zero-frequencycomponent (Av=0) and a frequency-dependent component (Av>0). The overall interdroplet pairpotential is therefore given by:

w h w h w hv vVDW ( ) ( ) ( )= +=0 0> (3.5)

where wv=0(h) and wv>0(h) are determined by inserting the expressions for Av=0 and Av>0 intoEquation 3.2 or 3.3. The zero-frequency component is due to orientation and inductioncontributions to the van der Waals interaction, whereas the frequency-dependent componentis due to the dispersion contribution (Section 2.3.3). The separation of the Hamaker functioninto these two components is particularly useful for understanding the influence of electro-static screening and retardation on van der Waals interactions (see below).

For food emulsions, the Hamaker function is typically about 0.75 × 10–20 J, with about 42%of this coming from the zero-frequency contribution and 58% from the frequency-dependentcontribution. The physicochemical properties needed to calculate Hamaker functions foringredients typically found in food emulsions are summarized in Table 3.1. In practice, themagnitude of the Hamaker function depends on droplet separation and is considerablyoverestimated by Equation 3.4 because of the effects of electrostatic screening, retardation,and interfacial layers (see below).

3.3.4. Electrostatic Screening

The zero-frequency component of the Hamaker function (Av=0) is electrostatic in originbecause it depends on interactions that involve permanent dipoles (Section 2.3.3). Conse-quently, this part of the van der Waals interaction is “screened” (reduced) when droplets are

TABLE 3.1Physicochemical Properties Needed to Calculate the NonretardedHamaker Function (Equation 3.4) for Some Materials CommonlyFound in Food Emulsions

Static relative Absorptiondielectric constant Refractive index frequency

Medium (R) (n) (ve/1015 s–1)

Water 80 1.333 3.0Oil 2 1.433 2.9Pure protein 5 1.56 2.9x% protein in water 5x + 80(1 – x) 1.56x + 1.333(1 – x) 2.9Pure Tween 20 1.468 2.9

Data compiled from Israelachvili (1992), Wei et al. (1994), and Hato et al. (1996).


suspended in an electrolyte solution because of the accumulation of counterions around thedroplets (Section 3.4.2). Electrostatic screening causes the zero-frequency component todecrease with increasing droplet separation and with increasing electrolyte concentration(Mahanty and Ninham 1976; Marra 1986; Israelachvili 1992; Mishchuk et al. 1995, 1996).At high electrolyte concentrations, the zero-frequency contribution decays rapidly with dis-tance and makes a negligible contribution to the overall interaction energy at distances greaterthan a few nanometers (Figure 3.3). On the other hand, the frequency-dependent component(Av>0) is unaffected by electrostatic screening because the ions in the electrolyte solution areso large that they do not have time to move in response to the rapidly fluctuating dipoles(Israelachvili 1992). Consequently, the van der Waals interaction may decrease by as much

FIGURE 3.3 Influence of electrostatic retardation and screening on van der Waals interactionsbetween oil droplets suspended in water (see Table 3.1 for the properties of the phases used in thecalculations). The curves in Figure 3.3a are for different electrolyte concentrations.

(a)

(b)


as 42% in oil-in-water emulsions at high-ionic-strength solutions because the zero-frequencycomponent is completely screened.

3.3.5. Retardation

The strength of the van der Waals interaction between emulsion droplets is reduced becauseof a phenomenon known as retardation (Israelachvili 1992). The origin of retardation is thefinite time taken for an electromagnetic field to travel from one droplet to another and back(Mahanty and Ninham 1976). The frequency-dependent contribution to the van der Waalsinteraction (wv>0) is the result of a transient dipole in one droplet inducing a dipole in anotherdroplet, which then interacts with the first dipole (Section 2.3.3). The strength of the resultingattractive force is reduced if the time taken for the electromagnetic field to travel between thedroplets is comparable to the lifetime of a transient dipole, because then the orientation of thefirst dipole will have changed by the time the field from the second dipole arrives (Israelachvili1992). This effect becomes appreciable at dipole separations greater than a few nanometersand results in a decrease in the frequency-dependent (Av>0) contribution to the Hamakerfunction with droplet separation. The zero-frequency contribution (Av=0) is unaffected byretardation because it is electrostatic in origin (Mahanty and Ninham 1976). Consequently,the contribution of the Av>0 term becomes increasingly small as the separation between thedroplets increases. For example, the retarded value of wv>0(h) between two emulsion dropletsat a separation of 20 nm is only 33% of the nonretarded value (Figure 3.3). Any accurateprediction of the van der Waals interaction between droplets should therefore include retar-dation effects. A number of authors have developed relatively simple correction functionswhich can be used to account for retardation effects (Schenkel and Kitchner 1960; Gregory1969, 1981; Anandarajah and Chen 1995; Chen and Anandarajah 1996), although the mostaccurate method is to solve the full theory numerically (Mahanty and Ninham 1976, Pailthorpeand Russel 1982).

3.3.6. Influence of an Interfacial Layer

So far, we have assumed that the van der Waals interaction occurs between two homogeneousspheres separated by an intervening medium (Figure 3.1). In reality, emulsion droplets arenormally surrounded by a thin layer of emulsifier molecules, and this interfacial layer hasdifferent physicochemical properties (εR, n, and ve) than either the oil or water phases (Figure3.4). The molecules nearest the surface of a particle make the greatest contribution to theoverall van der Waals interaction, and so the presence of an interfacial layer can have a largeeffect on the interactions between emulsion droplets, especially at close separations (Vold1961, Israelachvili 1992, Parsegian 1993).

The influence of an adsorbed layer on the van der Waals interactions between emulsiondroplets has been considered by Vold (1961):

FIGURE 3.4 The droplets in food emulsions are normally surrounded by an adsorbed emulsifier layer,which modifies their van der Waals interactions.


w h

A Hh

rA H

hr

A Hh

rr

r

VDW ( )

( ), , ,

=

−+

+ +

+ + +

112 131 232 1322

12

21 2

2∂∂ ∂ ∂

(3.6)

where the subscripts 1, 2, and 3 refer to the continuous phase, droplet, and emulsifier layer,respectively; h is the surface-to-surface separation between the outer regions of the adsorbedlayers; δ is the thickness of the adsorbed layer; and H(x,y) is a function given by:

H x yy

x xy xy

x xy x yx xy x

x xy x y( , ) ln=

+ ++

+ + ++ + +

+ + +

2 2

2

22

The dependence of the (nonretarded and nonscreened) van der Waals interaction betweentwo emulsion droplets on the thickness and composition of an interfacial layer consisting ofa mixture of protein and water was calculated using Equation 3.6 and the physical propertieslisted in Table 3.1 (Figure 3.5). In the absence of the interfacial layer, the attraction betweenthe droplets was about –110 kT at a separation of 1 nm. Figure 3.5 clearly indicates that theinterfacial layer causes a significant alteration in the strength of the interactions between thedroplets, leading to either an increase or decrease in the strength of the attraction, dependingon the protein concentration. At high protein concentrations (>60%), the attraction is greaterthan that between two bare emulsion droplets, whereas at low protein concentrations (<60%)it is smaller.

FIGURE 3.5 Influence of the composition of an interfacial layer, consisting of water and protein, onthe van der Waals interactions between emulsion droplets. The interdroplet pair potential is reported atan outer surface-to-surface separation of 1 nm for 1-µm droplets. The physical properties of the oil,water, and interfacial layer used in the calculations are reported in Table 3.1


3.3.7. General Features of van der Waals Interactions

1. The interaction between two oil droplets (or between two water droplets) is alwaysattractive.

2. The strength of the interaction decreases with droplet separation, and the interac-tion is fairly long range (w ∝ 1/h).

3. The interaction becomes stronger as the droplet size increases.4. The strength of the interaction depends on the physical properties of the droplets

and the surrounding liquid (through the Hamaker function).5. The strength of the interaction depends on the thickness and composition of the

adsorbed emulsifier layer.6. The strength of the interaction decreases as the concentration of electrolyte in an

oil-in-water emulsion increases because of electrostatic screening.

van der Waals interactions act between all types of colloidal particles, and therefore theymust always be considered when calculating the overall interaction potential between emul-sion droplets (Hiemenz 1986, Israelachvili 1992). Nevertheless, it must be stressed that anaccurate calculation of their magnitude and range is extremely difficult, because of the lackof physicochemical data required to perform the calculations and because of the need tosimultaneously account for the effects of screening, retardation, and interfacial layers (Hunter1986). The fact that van der Waals interactions are relatively strong and long range, and thatthey are always attractive, suggests that emulsion droplets would tend to associate with eachother. In practice, many food emulsions are stable to droplet aggregation, which indicates theexistence of repulsive interactions that are strong enough to overcome the van der Waalsattraction. Some of the most important types of these repulsive interactions, including elec-trostatic, polymeric steric, hydration, and thermal fluctuation interactions, are discussed in thefollowing sections.

3.4. ELECTROSTATIC INTERACTIONS

3.4.1. Origins of Surface Charge

The droplets in many food emulsions have electrically charged surfaces because of theadsorption of emulsifiers which are either ionic or capable of being ionized (e.g., proteins,polysaccharides, and surfactants) (Chapter 4). All food proteins have acidic (–COOH →COO– + H+) and basic (NH2 + H+ → NH3

+) groups whose degree of ionization depends on thepH and ionic strength of the surrounding aqueous phase (Charalambous and Doxastakis 1989,Damodaran 1996, Magdassi 1996, Magdassi and Kamyshny 1996). Some surface-activepolysaccharides, such as modified starch and gum arabic, also have acidic groups which maybe ionized (BeMiller and Whistler 1996). Ionic surfactants may be either positively ornegatively charged depending on the nature of their hydrophilic head group (Linfield 1976,Myers 1988, Richmond 1990). The magnitude and sign of the electrical charge on anemulsion droplet therefore depend on the type of emulsifier used to stabilize it, the concen-tration of the emulsifier at the interface, and the prevailing environmental conditions (e.g.,pH, temperature, and ionic strength). All the droplets in an emulsion are usually stabilized bythe same type of emulsifier and therefore have the same electrical charge. The electrostaticinteraction between similarly charged droplets is repulsive, and so electrostatic interactionsplay a major role in preventing droplets from coming close enough together to aggregate.

It is convenient to divide the different types of ions which can influence surface charge intothree categories (Hunter 1986, 1989):


1. Potential-determining ions. This type of ion is responsible for the association–dissociation of charged groups (e.g., –COOH → COO– + H+). In food emulsions,the most important potential-determining ions are H+ and OH–, because theygovern the degree of ionization of acidic and basic groups on many proteins andpolysaccharides. The influence of potential-determining ions on surface charge istherefore determined principally through the pH of the surrounding solution.

2. Indifferent electrolyte ions. This type of ion accumulates around charged groupsbecause of electrostatic interactions (e.g., Na+ ions may accumulate around anegatively charged –COO– group). These ions reduce the strength of the electricalfield around a charged group principally due to electrostatic screening (Section3.4.2), rather than causing association–dissociation of the charged group. At highionic strengths, some “indifferent” electrolyte ions can actually alter the degreeof ionization of charged groups. They do this either by altering the dissociationconstant of the surface groups (i.e., their pK value) or by acting as potential-determining ions that compete with the H+ or OH– ions (e.g., –COO– + Na+ →–COO–Na+). The influence of indifferent electrolyte ions on surface charge istherefore determined principally by the ionic strength of the surrounding solution.

3. Adsorbed ions. Surface charge can also be altered by the adsorption of surface-active ions. In food emulsions, the most important types of surface-active ions areionic emulsifiers, including many surfactants, proteins, and polysaccharides (Chapter4). The contribution of adsorbed ions to surface charge is governed mainly by thetype and concentration of emulsifiers present in the system and their relativeaffinities for the droplet surface.

Emulsion scientists are interested in understanding the role that each of these differenttypes of ions play in determining the electrical charge on emulsion droplets, because themagnitude of this charge determines the stability of many food emulsions to aggregation andtherefore has a pronounced influence on their appearance, taste, texture, and stability (Chap-ters 7 to 9).

3.4.2. Ion Distribution Near a Charged Surface

An understanding of the origin and nature of electrostatic interactions between emulsiondroplets relies on an appreciation of the way that the various types of ions are organized closeto a charged surface. Consider a charged surface which is in contact with an electrolytesolution (Figure 3.6). Ions of opposite charge to the surface (counterions) are attracted towardit, whereas ions of similar charge (co-ions) are repelled from it. Nevertheless, the tendencyfor ions to be organized in the vicinity of a charged surface is opposed by the disorganizinginfluence of the thermal energy (Evans and Wennerstrom 1994). Consequently, the concen-tration of counterions is greatest at the charged surface and decreases as one moves awayfrom the surface until it reaches the bulk counterion concentration, whereas the concentrationof co-ions is smallest at the charged surface and increases as one moves away from thesurface until it reaches the bulk co-ion concentration (Figure 3.6). The concentration ofcounterions near a charged surface is always greater than the concentration of co-ions, andso a charged surface can be considered to be surrounded by a cloud of counterions. Never-theless, the overall system must be electrically neutral, and so the charge on the surface mustbe completely balanced by the excess charge of the counterions in the electrolyte solution.The distribution of ions close to a charged surface is referred to as the electrical double layer,because it is convenient to assume that the system consists of two oppositely charged layers:


the surface and the surrounding liquid (Kitakara and Watanabe 1984; Hunter 1986, 1989;Hiemenz 1986).

It is important to establish the relationship between the characteristics of a charged surface,the properties of the solution in contact with it, and the distribution of ions in its immediatevicinity, because this information is needed to calculate the strength of electrostatic interac-tions between emulsion droplets. A surface is usually characterized by its surface chargedensity (σ) and its surface potential (Ψ

0). The surface charge density is the amount of

electrical charge per unit surface area, whereas the surface potential is the amount of energyrequired to increase the surface charge density from zero to σ. These values depend on thetype and concentration of emulsifier present at a surface, as well as the nature of theelectrolyte solution (e.g., pH, ionic strength, and temperature). The important characteristicsof an electrolyte solution are its dielectric constant and the concentration and valency of theions it contains.

A mathematical relationship, known as the Poisson–Boltzmann equation, has been derivedto relate the electrical potential in the vicinity of a charged surface to the concentration andtype of ions present in the adjacent electrolyte solution (Evans and Wennerstrom 1994):

d x

dxe

z nz e x

kTRi i

i

i

2

20

0ψ

ε εψ( )

exp( )= − −

∑ (3.7)

where n0i is the concentration of ionic species of type i in the bulk electrolyte solution (in

molecules per cubic meter), zi is their valency, e is the electrical charge of a single proton,

ε0 is the dielectric constant of a vacuum, ε

R is the relative dielectric constant of the solution,

and ψ(x) is the electrical potential at a distance x from the charged surface. This equation isof central importance to emulsion science because it is the basis for the calculation ofelectrostatic interactions between emulsion droplets. Nevertheless, its widespread applicationhas been limited because it does not have an explicit analytical solution (Hunter 1986). Whenaccurate calculations are required, it is necessary to solve Equation 3.7 numerically using adigital computer (Carnie et al. 1994). For certain systems, it is possible to derive muchsimpler analytic formulas which can be used to calculate the electrical potential near a surface

FIGURE 3.6 The organization of ions near a charged surface is governed by two opposing tendencies:(1) electrostatic interactions which favor accumulation of counterions near a surface and (2) thermalenergy which favors a random distribution of the ions.


by making certain simplifying assumptions (Evans and Wennerstrom 1994, Sader et al.1995).

If it is assumed that the electrostatic attraction between the charged surface and thecounterions is relatively weak compared to the thermal energy (i.e., z

ieψ0 < kT, which means

that ψ0 must be less than about 25 mV at room temperature in water), then a simpleexpression, known as the Debye–Huckel approximation, can be used to calculate the depen-dence of the electrical potential on distance from the surface (Hunter 1986, Hiemenz 1986):

Ψ Ψ( ) exp( )x x= −0 κ (3.8)

This equation indicates that the electrical potential decreases exponentially with distancefrom the surface at a decay rate which is determined by the parameter κ –1, which is knownas the Debye screening length. The Debye screening length is a measure of the “thickness”of the electrical double layer and it is related to the properties of the electrolyte solution bythe following equation:

κε ε− =∑

1 0

20

2

R

i i

kT

e n z (3.9)

For aqueous solutions at room temperatures, κ –1 ≈ 0.304/√I nm, where I is the ionicstrength expressed in moles per liter (Israelachvili 1992). For example, the Debye screeninglengths for NaCl solutions with different ionic strengths are 0.3 nm for a 1 M solution, 0.96nm for a 100 mM solution, 3 nm for a 10 mM solution, 9.6 nm for a 1 mM solution, and 30.4nm for a 0.1 mM solution.

The Debye screening length is an extremely important characteristic of an electrolytesolution because it determines how rapidly the electrical potential decreases with distancefrom the surface. Physically, κ –1 corresponds to the distance from the charged surface wherethe electrical potential has fallen to 1/e of its value at the surface. This distance is particularlysensitive to the concentration and valency of the ions in an electrolyte solution. As the ionconcentration or valency increases, κ –1 becomes smaller, and therefore the electrical potentialdecreases more rapidly with distance (Figure 3.7). The physical explanation for this phenom-enon is that the neutralization of the surface charge occurs at shorter distances when theconcentration of opposite charge in the surrounding solution increases.

A charged surface can be considered to be surrounded by a “cloud” of counterions witha thickness equal to the Debye screening length and which depends strongly on the ionconcentration and valency (Hunter 1986). As will be seen in later chapters, the screening ofelectrostatic interactions by electrolytes has important consequences for the stability andrheology of many food emulsions (Chapters 7 and 8).

The Poisson–Boltzmann theory assumes an electrolyte solution is a continuum that con-tains ions which are infinitesimally small. It therefore allows ions to accumulate at anunphysically large concentration near to a charged surface (Evans and Wennerstrom 1994,Kitakara and Watanabe 1984). In reality, ions have a finite size and shape, and this limitsthe number of them which can be present in the first layer of molecules that are in directcontact with the surface (Derjaguin et al. 1987, Derjaguin 1989, Israelachvili 1992). Thisassumption is not particularly limiting for systems in which there is a weak interactionbetween the ions and the charged surface. Nevertheless, it becomes increasingly unrealisticas the strength of the electrostatic interactions between a charged surface and the surround-


FIGURE 3.7 Influence of ionic strength on the electric field near a charged surface. The electricaldouble layer shrinks as the electrolyte concentration increases or the ion valency increases.

ing ions increases relative to the thermal energy (Evans and Wennerstrom 1994). In thesecases, the Poisson–Boltzmann theory must be modified to take into account the finite sizeof the ions in the electrolyte solution. It has proved convenient to divide the counteriondistribution near a highly charged surface into two regions: an inner and an outer region(Figure 3.8).

3.4.2.1. Inner Region

In the inner region, the attraction between the counterions and charged surface is strong andtherefore they are relatively immobile, whereas in the outer region, the attraction is muchweaker and therefore the counterions are more mobile (Evans and Wennerstrom 1994). Thethickness of the inner region (δ) is approximately equal to the radius of the hydratedcounterions, rather than their diameter, because the effective charge of an ion is located at itscenter (Hiemenz 1986). The inner region is sometimes referred to as the Stern layer, whilethe boundary between the inner and outer regions is referred to as the Stern plane (Figure3.8), after Otto Stern, the scientist who first proposed this concept (Hiemenz 1986, Derjaguin1989). The electrical potential at the Stern plane (Ψδ) is different from that at the surface (Ψ0)because of the presence of the counterions in the Stern layer. For monovalent indifferentelectrolyte counterions, Ψδ is less than Ψ0, because the surface charge is partly neutralizedby the charge on the counterions (Figure 3.9a). The extent of this decrease depends on thenumber and packing of the counterions within the Stern plane (Derjaguin 1989). The samebehavior is observed for multivalent indifferent electrolyte counterions at low concentrations,but at higher concentrations the surface may adsorb such a large number of oppositelycharged multivalent counterions that its charge is actually reversed, so that Ψδ has an oppositesign to Ψ0 (Figure 3.9b). If a charged surface adsorbs surface-active co-ions (e.g., ionicemulsifiers), it is even possible for Ψδ to be larger than Ψ0 (Figure 3.9c). An increase insurface charge may occur when the hydrophobic attraction between the nonpolar tail of asurfactant and a surface is greater than any electrostatic repulsive interactions. Thus themagnitude of the electrical potential at the Stern plane depends on the precise nature andconcentration of the ions present in the system (Derjaguin 1989).


FIGURE 3.9 The electrical potential at the Stern plane may be lower, higher, or a different sign thanthat at the surface, depending on the strength of the interaction and the type of ions adsorbed.

A number of theories have been developed to take into account the effect of the finite sizeand limited packing of ions in the Stern layer on the relationship between Ψδ and Ψ0. Oneof the most widely used is the Stern isotherm, which assumes that there are only a finitenumber of binding sites at the surface and that once these are filled, the surface becomessaturated and cannot adsorb any more ions:

θχ

χ=

+K

K1 (3.10)

where θ is the fraction of occupied surface sites, K is is the adsorption equilibrium constant,and χ is the mole fraction of the ion in the bulk phase. The adsorption equilibrium constantdepends on the strength of the interaction between the ion and the surface compared to thethermal energy (Hiemenz 1986):

KG

kT≈

exp

∆ ads(3.11)

FIGURE 3.8 When the electrostatic attraction between a charged surface and the surroundingcounterions is relatively strong compared to the thermal energy, it is convenient to divide the electrolytesolution into an inner and an outer region.

(a) (b) (c)


where ∆Gads is the free energy associated with the adsorption of an ion to the interface: ∆Gads

= zeΨδ + φ. The zeΨδ term is due to the electrostatic attraction between the ion and thesurface, while the φ term accounts for any specific binding effects. These specific bindingeffects could be due to hydrophobic interactions (e.g., when an emulsifier adsorbs) or chemi-cal interactions (e.g., –COO– + Na+ → –COO–Na+). The fraction of surface sites which areoccupied increases as the free energy of adsorption of an ion increases. The electricalpotential at the Stern layer can be related to the electrical potential at the charged surfaceusing the following equation (Hiemenz 1986):

ψ ψ δθσε εδ

δ= −0

0

*(3.12)

where σ* is the surface charge density when the surface is completely saturated with ions andεδ is the relative dielectric constant of the Stern layer. Equation 3.12 indicates that thedifference between the potential at the surface and that at the Stern plane depends on thefraction of surface sites that are occupied. In principle, this equation can be used to calculatethe change in the electrical potential of a surface due to ion adsorption. In practice, thisequation is difficult to use because of a lack of knowledge about the values of δ, φ, and εδin the Stern layer (Hiemenz 1986, Derjaguin 1989). These parameters are unique for everyion–surface combination and are difficult to measure experimentally. For this reason, it isusually more convenient to experimentally measure the electrical potential at the Stern plane,rather than attempting to predict it theoretically (Hunter 1986).

Experiments have shown that Ψδ is closely related to the electrical potential at the shearplane (Hunter 1986). When a liquid flows past a charged surface, it “pulls” those counterionswhich are only weakly attached to the surface along with it, but leaves those ions that arestrongly attached in place (i.e., those ions in the Stern plane). The shear plane is defined asthe distance from the charged surface below which the counterions remain strongly attachedand is approximately equal to the diameter of the hydrated ions (Figure 3.8). The electricalpotential at the shear plane is referred to as the zeta potential (ζ) and can be measured usingvarious types of electrokinetic techniques (Chapter 10).

3.4.2.2. Outer Region

In the outer region, the electrostatic interaction between the surface and the counterions isusually fairly weak (because the ions in the Stern layer partially screen the surface charge),and so the variation in electrical potential with distance can be described by Equation 3.8, byreplacing Ψ0 with Ψδ:

Ψ Ψ( ) exp( )x x= −δ κ (3.13)

where x is now taken to be the distance from the shear plane, rather than from the chargedsurface. The dependence of the electrical potential on distance from the shear plane can thenbe calculated once Ψδ is known.

3.4.3. Electrostatic Interactions Between Charged Droplets

In food emulsions, we are interested in the strength of the electrostatic interactions betweendroplets (Dickinson and Stainsby 1982, Dickinson 1992). An isolated charged droplet issurrounded by a cloud of counterions with a thickness determined by the Debye screening


length (Figure 3.10). When two similarly charged droplets approach each other, their counterionclouds overlap, and this gives rise to a repulsive interaction (Evans and Wennerstrom 1994).There are two major contributions to this electrostatic interaction: (1) an enthalpic contribu-tion associated with the change in the strength of the attractive and repulsive electrostaticinteractions between the various charged species involved and (2) an entropic contributionassociated with the confinement of the counterions between the droplets to a smaller volume.The entropic contribution is strongly repulsive, whereas the enthalpic contribution is weaklyattractive, and therefore the overall interaction is repulsive (Evans and Wennerstrom 1994).The fact that the major contribution to the electrostatic interaction is entropic means that itincreases in strength with increasing temperature.

Theoretical equations based on the Poisson–Boltzmann theory have been derived to relatethe electrostatic interdroplet pair potential to the physical characteristics of the emulsiondroplets and the intervening electrolyte solution (Hiemenz 1986, Hunter 1986, Carnie et al.1994, Okshima 1994, Sader et al. 1995). The full equations cannot be solved analytically, butthey can be solved numerically on a computer (Carnie et al. 1994) or by making simplifyingassumptions that lead to relatively simple equations that are applicable under certain condi-tions (Sader et al. 1995).

If it is assumed that there is a relatively low surface potential (Ψδ < 25 mV) and that theDebye screening length and surface-to-surface separation are much less than the droplet size(i.e., κ –1 < r /10 and h < r /10), then fairly simple expressions for the electrostatic interdropletpair potential between two similar droplets can be derived (Hunter 1986).

At constant surface potential:

w h r hRelectrostaticψ

δπε ε ψ κ( ) ln[ exp( )]= + −2 102 (3.14)

At constant surface charge:

w h r hRelectrostaticσ

δπε ε ψ κ( ) ln[ exp( )]= − − −2 102 (3.15)

The smallest droplets in most food emulsions are about 0.1 µm in radius, which means thatthese equations are likely to be applicable at droplet separations less than about 10 nm andat electrolyte concentrations greater than about 1 mM. Whether the electrostatic interactionbetween two droplets takes place under conditions of constant surface potential or constantsurface charge depends on the ability of the surface groups to regulate their charge (Israelachvili1992).

FIGURE 3.10 Distribution of ions around electrically charged emulsion droplets.


So far, it has been assumed that the charge on the droplets is evenly spread out over thewhole of the surface. In practice, droplets may have surfaces which have some regions whichare negatively charged, some regions which are positively charged, and some regions whichare neutral. The heterogeneous distribution of the charges on a droplet may influence theirelectrostatic interactions (Holt and Chan 1997). Thus, two droplets (or molecules) which haveno net charge may still be electrostatically attracted to each other if they have patches ofpositive and negative charge.

3.4.3.1. Charge Regulation

As two similarly charged emulsion droplets move closer together, the interaction betweenthem becomes increasingly repulsive. Certain systems are capable of reducing the magnitudeof this increase by undergoing structural rearrangements, which is referred to as chargeregulation. For example, the surface charge may be regulated by adsorption–desorption ofionic emulsifiers (Yaminsky et al. 1996a,b) or by association–dissociation of charged groups(Hunter 1986, 1989). Depending on the physical characteristics of a system, it is possible todiscern three different situations which may occur when two droplets approach each other(Reiner and Radke 1993):

1. Constant surface charge. As the droplets move closer together, the number ofcharges per unit surface area remains constant (i.e., no adsorption–desorption orassociation–dissociation of ions occurs). In this case, the electrostatic repulsionbetween the surfaces is at the maximum possible value because the surfaces arefully charged.

2. Constant surface potential. As the droplets move closer together, the number ofcharges per unit surface area decreases (e.g., by an adsorption–desorption orassociation–dissociation mechanism). In this case, the electrostatic repulsion be-tween the surfaces is at the minimum possible value because the surface charge isreduced.

3. Charge regulation. In reality, the electrostatic repulsion usually falls somewherebetween the two extremes mentioned above because of charge regulation. Thenumber of charges per unit surface area depends on the characteristics of theadsorption–desorption or association–dissociation mechanisms (e.g., the surfaceactivity of an ionic emulsifier or the surface pK value of an ionizable group). Theseprocesses take a finite time to occur, and therefore the surface charge density mayalso depend on the speed at which the droplets come together (Israelachvili 1992,Israelachvili and Berman 1995).

The variation of the interdroplet pair potential with separation is shown for two similarlycharged droplets in Figure 3.11. There is a strong repulsive interaction between the dropletsat close separations, which decreases as the droplets move farther apart. This repulsiveinteraction is often sufficiently strong and long range to prevent droplets from aggregating.At relatively large droplet separations, Equations 3.14 and 3.15 give approximately the samepredictions for the electrostatic interaction, but at closer separations, the assumption ofconstant charge predicts a significantly higher repulsion than the assumption of constantpotential (Figure 3.11). In practice, the interdroplet pair potential always lies somewherebetween these two extremes and depends on the precise nature of the system.

The magnitude and range of the electrostatic repulsion between two droplets decrease asthe ionic strength of the solution separating them increases because of electrostatic screening(i.e., the accumulation of counterions around the surfaces) (Figure 3.12). This has important


FIGURE 3.12 Electrolyte reduces the magnitude and range of the electrostatic repulsion betweenemulsion droplets due to electrostatic screening.

FIGURE 3.11 Comparison of electrostatic interaction between a pair of emulsion droplets underconditions of constant surface charge and constant surface potential.

consequences for the texture and stability of many food emulsions and explains the suscep-tibility of protein-stabilized emulsions to flocculation when the electrolyte concentration isincreased above a critical level (Demetriades et al. 1997a).

3.4.3.2. Effect of Electrolyte on Surface Potential

When the electrostatic interaction between a charged surface and the counterions is relativelyweak, the surface charge density is simply related to the surface potential: σ = εRε0κΨδ. Thisequation indicates that the electrical properties of a surface are altered by the presence ofelectrolytes in the aqueous phase and has important consequences for the calculation of the


FIGURE 3.13 Polyvalent ions are capable of forming ion bridges between emulsion droplets.

electrostatic interdroplet pair potential. If the surface charge density remains constant whensalt is added to the aqueous phase, then the surface potential decreases (because less energyis needed to bring a charge from infinity to the droplet surface through an electrolytesolution). Conversely, if the electrical potential remains constant as the salt concentration isincreased, this means that the surface charge density must decrease. In practice, both σ andΨδ tend to change simultaneously. In food emulsions, one can usually assume that the surfacecharge density is independent of ionic strength at low to moderate electrolyte concentrations,and so one must take into account the variation in Ψδ with ionic strength when calculatingthe electrostatic repulsion.

3.4.4. Ion Bridging

Ion bridging is another type of colloidal interaction which involves electrostatic interactions(Ducker and Pashley 1992). It occurs when a polyvalent ion simultaneously binds to thesurface of two emulsion droplets that have an opposite charge to the ion (Figure 3.13). Thesepolyvalent ions may be low-molecular-weight species, such as Ca2+, Mg2+, or Al3+ (Dickinsonet al. 1992; Agboola and Dalgleish 1995, 1996), or high-molecular-weight biopolymers, suchas polysaccharides or proteins. The tendency for ion bridging to occur depends on thestrength of the ion bridge that holds the droplets together compared to the electrostaticrepulsion between the similarly charged droplets. For this reason, large polyvalent species,such as ionic polysaccharides, are often most effective at forming ion bridges because theyare able to act as a bridge between the droplets without allowing them to get too closetogether. The ability of polyvalent ions to form ion bridges is superimposed on their abilityto reduce the electrostatic repulsion between droplets through charge screening.

3.4.5. General Features of Electrostatic Interactions

1. Electrostatic interactions may be either attractive or repulsive depending on thesign of the charges on the droplets. The interaction is repulsive when droplets havesimilar charges (which is usually the case), but is attractive when they haveopposite charges.

2. The strength of the interaction decreases with droplet separation and may be eitherlong or short range depending on the ionic strength of the electrolyte solutionsurrounding the droplets. The interaction becomes increasingly short range as theionic strength increases because of electrostatic screening.

3. The strength of the interaction is proportional to the size of the emulsion droplets.4. The strength of the interaction depends on the electrical characteristics of the

droplet surfaces (e.g., the number of emulsifier molecules adsorbed per unit sur-face area, the number of ionizable groups per emulsifier molecule, and the concen-tration of any potential-determining ions in the aqueous phase, for example, H+ orOH–).


5. The interaction becomes more difficult to predict when association–dissociation ofionizable groups or adsorption–desorption of ionic emulsifiers occurs, especially atclose droplet separations.

6. Ion bridging effects have to be taken into account when polyvalent ions areinvolved.

In this section, we have seen that under certain conditions repulsive electrostatic interac-tions may be relatively strong and long range compared to attractive van der Waals interac-tions (compare Figures 3.3 and 3.11). This suggests that they may be strong enough toprevent droplets from aggregating in certain systems. Indeed, it is widely recognized thatelectrostatic stabilization plays an important role in determining the aggregation of dropletsin many food emulsions and particularly those stabilized by proteins (Friberg and Larsson1997, Dickinson and Stainsby 1982, Dickinson 1992). It should also be recognized thatelectrostatic interactions influence various other properties of food emulsions, such as thepartitioning of ingredients and the rates of chemical reactions. For example, the partitioningof an ionizable volatile flavor compound, such as butyric acid, between the head space andbulk of an emulsion is influenced by electrostatic interactions (Guyot et al. 1996). Thenegatively charged flavor component is attracted to positively charged droplets, which causesits volatility to be decreased, thus reducing the aroma. Lipid oxidation in food emulsions isoften catalyzed by polyvalent ions, such as Fe3+, that are normally present in the aqueousphase. The rate of iron-catalyzed lipid oxidation in oil-in-water emulsions has been shown toincrease when the droplets have a negative charge because the Fe3+ catalyst and oil moleculesare brought into close contact (Coupland and McClements 1996, Mei et al. 1998). A knowl-edge of the factors which determine the magnitude and range of electrostatic interactions istherefore extremely important to food scientists.

3.5. POLYMERIC STERIC INTERACTIONS

3.5.1. Polymeric Emulsifiers

In Section 3.3, we saw that van der Waals interactions always operate between emulsiondroplets and that these interactions are strong enough to cause droplets to aggregate, unlessthere is a sufficiently strong repulsive interaction to prevent them from coming close together.When emulsion droplets are surrounded by a layer of electrically charged emulsifier mol-ecules, they may be stabilized against aggregation by electrostatic repulsion (Section 3.4).Nevertheless, many food emulsions are stable to droplet aggregation, despite being sur-rounded by a layer of emulsifier molecules that has no electrical charge, which indicates thatother types of repulsive interaction also play an important role in stabilizing these systems(Dickinson and Stainsby 1982, Dickinson 1992). The most important of these is polymericsteric stabilization, which is due to the presence of polymeric emulsifiers at the oil–waterinterface. To be effective at preventing droplet aggregation, steric interactions must becomparable, both in magnitude and range (but opposite in sign), to van der Waals and anyother attractive interactions.

Most emulsifiers used in the food industry are either partly or entirely polymeric: proteinsare polymers of amino acids, polysaccharides are polymers of sugars, and many nonionicsurfactants have polar head groups which are polymers of oxyethylene (Chapter 4). Theconformation of a polymeric emulsifier at an interface depends on the number, type, andsequence of monomers along its backbone (Dalgleish 1989, 1995, 1996a,b; Damodaran 1989,1990, 1996; Dickinson and McClements 1995). An improved understanding of the relation-ship between the interfacial conformation of polymeric emulsifiers and their ability to stabi-


(a) (b) (c)

lize emulsions is particularly important for food scientists because it enables them to createand select ingredients in a more systematic fashion.

An emulsifier tends to adopt an interfacial conformation which minimizes the free energyof the system (Damodaran 1989). The conformational free energy is determined by variousenthalpic (intermolecular interaction energies) and entropic (configurational entropy andsolvation) contributions. In practice, the major factor which determines the interfacial con-formation of food emulsifiers is the hydrophobic effect (i.e., the tendency for a molecule toadopt an arrangement which minimizes the number of unfavorable contacts between polarand nonpolar groups) (Damodaran 1989, Dickinson and McClements 1995). Thus, small-molecule surfactants adopt a conformation in which the hydrocarbon tails protrude into theoil phase, while the hydrophilic head groups protrude into the aqueous phase (Figure 3.14a).Flexible biopolymers, such as casein or modified starches, exist as a series of loops, tails, andtrains at an interface (Figure 3.14b). The predominantly hydrophilic or hydrophobic segmentswhich protrude into the aqueous or oil phases (respectively) are referred to as loops or tails,whereas the predominantly neutral segments that lie flat against the interface are referred toas trains (Dickinson 1992). Immediately after adsorption, compact biopolymers (such asglobular proteins) tend to adopt an orientation which maximizes the contact area between anyhydrophobic patches on their surface and the oil phase (Damodaran 1989). After adsorption,the protein molecules may change their conformation in response to their new environment(Dickinson and McClements 1995; Corredig and Dalgleish 1995; Dalgleish 1996a,b). It mustbe stressed that many biopolymers take an appreciable time to undergo conformationalchanges, and therefore the interface may not be at thermodynamic equilibrium (which isassumed in many theories describing their behavior).


Polymeric steric interactions arise when emulsion droplets get so close together that theemulsifier layers overlap (Figure 3.15). This type of interaction can be conveniently dividedinto two contributions (Hiemenz 1986, Hunter 1986):

w h w h w hsteric elastic mix( ) ( ) ( )= + (3.16)

The elastic contribution is due to the compression of the interfacial membrane, whereas themixing contribution is due to the intermingling of the polymer chains (Figure 3.15).

3.5.2.1. Mixing Contribution

If it is assumed that the polymer molecules in the layers interpenetrate each other without thelayers being compressed (Figure 3.15a), then the interaction is entirely due to mixing of the

FIGURE 3.14 Orientation of some polymeric emulsifiers at an oil–water interface: (a) small-moleculesurfactants, (b) flexible biopolymers, and (c) globular biopolymers.


(a)

(b)

polymers. The theories describing polymeric steric interactions are much less well developedthan those describing electrostatic or van der Waals interactions. The major reason for thisis that polymeric steric interactions are particularly sensitive to the precise structure, orien-tation, packing, and interactions of the polymer molecules at the interface (Hunter 1986,Claesson et al. 1995). These parameters vary from system to system and are difficult toaccount for theoretically or to measure experimentally. Mathematical theories have beendeveloped for a number of simple well-defined systems, and it is informative to examinethese because they provide some useful insights into more complex systems (Hunter 1986).For example, the following equation has been derived to account for the mixing contributionwhen the polymer molecules are permanently attached to the droplet surface and there is aconstant number of polymer chains per unit surface area (Hunter 1986):

w h rkTm Nv

VP

S

hmix A( ) = −

−

4 12

2 212

12

π χ δ (3.17)

where m is the mass of polymer chains per unit area, δ is the thickness of the adsorbed layer,NA is Avogadro’s number, χ is the Flory–Huggins parameter, vP is the partial specificvolume of the polymer chains, and VS is the molar volume of the solvent. The Flory–Hugginsparameter depends on the relative magnitude of the solvent–solvent, solvent–segment, andsegment–segment interactions and is a measure of the quality of a solvent. It is related to theeffective interaction parameter (w) which was introduced in Chapter 2 to characterize thecompatibility of molecules in mixtures: χ = w/RT. In a good solvent (χ < 0.5), the polymermolecules prefer to be surrounded by solvent molecules. In a poor solvent (χ > 0.5), thepolymer molecules prefer to be surrounded by each other. In an indifferent (theta) solvent (χ= 0.5), the polymer molecules have no preference for either solvent or polymer molecules.In the original Flory–Huggins theory, it was assumed that χ was entirely due to enthalpiccontributions associated with the molecular interactions. In practice, it is more convenient toassume that χ also contains entropic contributions since interactions involving changes in thestructural organization of the solvent can then be accounted for (e.g., hydrophobic interac-tions) (Evans and Wennerstrom 1994). Whether the mixing contribution is attractive orrepulsive depends on the quality of the solvent. In a good solvent, the increase in concentra-tion of polymer molecules in the interpenetration zone is thermodynamically unfavorable

FIGURE 3.15 Steric interactions between emulsion droplets can be divided into an elastic contribu-tion which involves compression of the polymer layers and a mixing contribution which involvesinterpenetration of the polymer chains.


(wmix positive) because it reduces the number of polymer–solvent contacts and therefore leadsto a repulsive interaction between the droplets. Conversely, in a poor solvent, it is thermo-dynamically favorable (wmix negative) because it increases the number of polymer–polymercontacts and therefore leads to an attractive interaction between the droplets. In an indifferentsolvent, the polymer molecules have no preference as to whether they are surrounded bysolvent or by other polymer molecules, and therefore the mixing contribution is zero. Thus,by altering solvent quality, it is possible to change the mixing contribution from attractive torepulsive or vice versa. In food emulsions, this could be done by varying temperature or byadding alcohol or electrolyte to the aqueous phase.

3.5.2.2. Elastic Contribution

If it is assumed that the polymer layers surrounding the emulsion droplets are compressedwithout any interpenetration of the polymer molecules (Figure 3.15b), then the interaction isentirely elastic. When the layers are compressed, a smaller volume is available to the polymermolecules and therefore their configurational entropy is reduced, which is energeticallyunfavorable, and so this type of interaction is always repulsive (welastic positive).

The magnitude of the elastic contribution can be calculated from a statistical analysis ofthe number of configurations the polymer chains can adopt before and after the layers arecompressed (Hiemenz 1986, Dickinson 1992):

w h kTvhelastic( ) ln

( )( )

= ∞2

ΩΩ (3.18)

where Ω(h) and Ω(∞) are the number of configurations available to the chains at a separationof h (compressed) and of infinity (uncompressed), respectively. In practice, it is difficult tocalculate the number of conformations that a polymer molecule can adopt, and therefore moreempirical methods have been developed to account for the elastic interaction. One of the mostconvenient to use was derived by Jackel (1964):

w h E h r h

w h h

elastic

elastic

( ) . ( )

( ) ( )

/

= −

+( ) <

= ≥

0 77

0

12

12

5 2

δ δ δ

δ(3.19)

where E is the elastic modulus of the adsorbed layer. This equation indicates that there is anegligible interaction between the droplets when the separation is greater than the thicknessof the emulsifier layers, but that there is a steep increase in the repulsive interaction energywhen the droplets approach closer than this distance (Figure 3.16). As a first approximation,it may be possible to use measurements of the elastic modulus of macroscopic solutions andgels with polymer concentrations similar to those found in the interfacial region in Equation3.19. Alternatively, the elastic modulus of an interfacial layer could be measured directlyusing various types of surface force apparatus (Claesson et al. 1995).

3.5.3. Distance Dependence of Polymeric Steric Interactions

Steric interactions between emulsion droplets are conveniently divided into three regimes,according to the separation of the surface of the bare droplets (h) relative to the thickness ofthe polymer layers (δ):


1. Zero interaction regime (h 2). At sufficiently large droplet separations, thepolymer layers do not overlap with each other and the steric interaction betweenthe droplets is zero.

2. Interpenetration regime ( h < 2). When the droplets are sufficiently closetogether for the polymer layer on one droplet to interpenetrate that on anotherdroplet, without significantly compressing it, the major contribution to the stericinteraction is the mixing contribution (wmix), which may be either positive ornegative depending on the quality of the solvent

3. Interpenetration and compression regime (h < ). When the droplets get soclose together that the polymer layers start to compress each other, the overallsteric interaction is a combination of elastic and mixing contributions, although thestrongly repulsive elastic component usually dominates, and so the overall inter-action is repulsive.

It should be stressed that the length of the interpenetration and elastic regions actuallydepends on the precise nature of the polymer molecules present at the interface (Claesson etal. 1995). Flexible biopolymer molecules will have relatively large interpenetration regions,whereas compact globular proteins will have relatively small ones. Consequently, the choiceof δ as the distance where the elastic contribution first contributes to the interaction isarbitrary, and different values will be more appropriate for some systems. As mentionedearlier, the only way these values can accurately be established for a particular system is bymeasuring the force between two polymer-coated surfaces as they are brought closer together(Israelachvili 1992, Claesson et al. 1995, 1996).

3.5.4. Optimum Characteristics of Polymeric Emulsifier

To be effective at providing steric stabilization, a polymeric emulsifier must have certainphysicochemical characteristics (Hunter 1986, Dickinson 1992). First, it must have some

FIGURE 3.16 Interdroplet pair potential due to steric polymeric interactions. At intermediate sepa-rations, the steric polymeric interaction can be either attractive or repulsive depending on the qualityof the solvent because of the mixing contribution, but at short separations it is strongly repulsivebecause of the elastic contribution.


segments which bind strongly to the droplet surface (to anchor the polymer to the surface)and other segments which protrude a significant distance into the surrounding liquid (toprevent the droplets from coming close together). This means that the emulsifier must beamphiphilic, with some hydrophobic segments which protrude into the oil phase and somehydrophilic segments which protrude into the aqueous phase (Figure 3.14). The binding tothe interface must be strong enough to prevent the emulsifier from desorbing from the dropletsurface as the droplets approach one another. Second, the continuous phase surrounding thedroplets must be a sufficiently good solvent for the segments which protrude into it, so thatthe mixing contribution to the overall interaction energy is repulsive (wmix positive). Third,the polymeric-repulsive interaction must act over a distance that is comparable to the rangeof the attractive van der Waals interactions. Thus uncharged biopolymers that form thickinterfacial layers, such as modified starch, are much more effective at stabilizing emulsionsagainst flocculation than biopolymers that form thin layers, such as globular proteins at theirisoelectric point.* Finally, the surface must be covered by a sufficiently high concentrationof polymer. If too little polymer is present, a single polymer may adsorb to the surface of twodifferent emulsion droplets, forming a bridge which causes the droplets to flocculate. Inaddition, some of the nonpolar regions will be exposed to the aqueous phase, which leads toa hydrophobic attraction between the droplets (see Section 3.7). Many polymeric emulsifiersare charged, and therefore they stabilize emulsion droplets against aggregation through acombination of electrostatic and steric repulsion (Claesson et al. 1995).

3.5.5. General Features of Polymeric Steric Stabilization

1. Interactions are always strongly repulsive at short separations (h < δ), but may beeither attractive or repulsive at intermediate separations (δ < h < 2δ) depending onthe quality of the solvent (Figure 3.16).

2. The range of the interaction increases with the thickness of the adsorbed layer.3. The strength of the interaction increases with droplet size.4. The strength of the interaction depends on the molecular architecture of the adsorbed

polymeric emulsifier layer and therefore varies considerably from system to sys-tem and is difficult to predict from first principles.

Polymer steric interactions are one of the most common and important stabilizing mecha-nisms in food emulsions. Unlike electrostatic interactions, they occur in almost every type offood emulsion because most emulsifiers are polymeric. Some emulsions are stabilized almostentirely by polymeric steric stabilization, whereas others are stabilized by a combination ofsteric and electrostatic stabilization. Food scientists must often decide which is the mostappropriate emulsifier for a particular application, and so it is useful to compare the differ-ences between steric and electrostatic stabilization (Table 3.2). The principal difference istheir sensitivity to pH and ionic strength (Hunter 1986). The electrostatic repulsion betweenemulsion droplets is dramatically decreased when the electrical charge on the droplet surfacesis reduced (e.g., by altering the pH) or screened (e.g., by increasing the concentration ofelectrolyte in the aqueous phase). In contrast, steric repulsion is fairly insensitive to bothelectrolyte concentration and pH.** Another major difference is the fact that the electrostaticrepulsion is usually weaker than the van der Waals attraction at short distances, whereas the

* Thick biopolymer layers may also help stabilize droplets against aggregation by reducing the magnitude of theattractive van der Waals interaction (Section 3.3.6).

** It should be stressed that the polymeric steric interaction may be affected by pH and ionic strength if the polymermolecules are charged, because this will alter the thickness of the interfacial layer and the interaction of thepolymer chains.


TABLE 3.2Comparison of the Advantages and Disadvantages of Polymericand Electrostatic Stabilization Mechanisms in Food Emulsions

Polymeric steric stabilization Electrostatic stabilization

1. Insensitive to pH pH dependent — aggregation tends to occur near the isoelectricpoint of biopolymers

2. Insensitive to electrolyte Aggregation tends to occur at high electrolyte concentrations3. Large amounts of emulsifier needed

to cover droplet surface Small amounts of emulsifier needed to cover droplet surface4. Weak flocculation (easily reversible) Strong flocculation (often irreversible)5. Good freeze–thaw stability Poor freeze–thaw stability

Adapted from Hunter 1986.

FIGURE 3.17 An attractive depletion interaction arises between emulsion droplets when they aresurrounded by small nonadsorbing colloidal particles.

polymeric steric stabilization is stronger (Hunter 1986). This means that droplets stabilizedby electrostatic repulsion are always thermodynamically unstable with respect to aggregation(because the aggregated state has a lower free energy than the unaggregated state), whereassterically stabilized systems may be thermodynamically stable if the thickness of the adsorbedlayer extends a sufficient distance into the continuous phase.

From a practical standpoint, another important difference is the fact that considerably moreemulsifier is usually required to provide steric stabilization (because a thick interfacial layeris required) than to provide electrostatic stabilization. Thus, >5% modified starch is requiredto stabilize a 20 wt% oil-in-water emulsion containing 1-µm droplets, whereas <0.5% wheyprotein is required to stabilize the same system. The amount of emulsifier required to preparean emulsion is often an important financial consideration when formulating a food product.

3.6. DEPLETION INTERACTIONS

3.6.1. Origin of Depletion Interactions

Many food emulsions contain small colloidal particles that are dispersed in the continuousphase which surrounds the droplets (Figure 3.17). These colloidal particles may be surfactant


micelles formed when the free surfactant concentration exceeds some critical value (Aronson1992, Bibette 1991, McClements 1994), individual polymer molecules (Sperry 1982, Seeberghand Berg 1994, Dickinson et al. 1995, Smith and Williams 1995, Jenkins and Snowden 1996),or aggregated polymers (Dickinson and Golding 1997a,b). The presence of these colloidalparticles causes an attractive interaction between the droplets which is often large enough topromote emulsion instability (Jenkins and Snowden 1996). The origin of this interaction isthe exclusion of colloidal particles from a narrow region surrounding each droplet (Figure3.17). This region extends a distance approximately equal to the radius (rc) of a colloidalparticle away from the droplet surface. The concentration of colloidal particles in this deple-tion zone is effectively zero, while it is finite in the surrounding continuous phase. As aconsequence, there is an osmotic potential difference which favors the movement of solventmolecules from the depletion zone into the bulk liquid, so as to dilute the colloidal particlesand thus reduce the concentration gradient. The only way this process can be achieved is bytwo droplets aggregating and thereby reducing the volume of the depletion zone, whichmanifests itself as an attractive force between the droplets (Figure 3.17). Thus there is anosmotic driving force that favors droplet aggregation and which increases as the concentra-tion of colloidal particles in the aqueous phase increases.


When the separation between two droplets is small compared to their size (h << rd), theinterdroplet pair potential due to exclusion of the colloidal particles from the depletion zoneis given by the following expression (Sperry 1982):

w h

r Pr

rhr

r

rhr

c c

depletion

OSM

( ) =

− +

+ +

− +

+

2

33

3 3 2

2 1 12

3 1 12

π (3.20)

where POSM is the osmotic pressure arising from the exclusion of the colloidal particles andrc is the radius of the colloidal particles. The osmotic pressure difference is given by thefollowing equation (Hiemenz 1986):

PCRTM

N CvOSM

A

2M= +

1 (3.21)

Here, C, M, and v are the concentration, molecular weight, and volume of the colloidalparticles, and NA is Avogadro’s number.

For homogeneous spherical particles, the equation for the interdroplet pair potential can bewritten more conveniently in the following form:

w h

kT rr

r

rhr

r

rhrc

c cc c

depletion( )

( )

=

−

+ +

+ +

− +

+

21 2 2 1 1

23 1 1

2

3 3 3 2

φ φ

(3.22)


where φc is the volume fraction of the colloidal particles. The dependence of the interdropletpair potential due to depletion effects on the droplet separation is shown in Figure 3.18. Theinteraction potential is zero at droplet separations greater than the diameter of the colloidalparticles and decreases to a constant value when the droplets come into close contact.

Depletion interactions rely on the colloidal particles not interacting strongly with thesurface of the emulsion droplets. Otherwise, the bound particles would have to be displacedas the droplets moved closer together, which would require the input of energy and thereforebe repulsive.

3.6.3. General Features of Depletion Interactions

1. The strength of the interaction increases as the concentration of colloidal particlesin the continuous phase increases (at constant rc).

2. The strength of the interaction increases as the size of the droplets in an emulsionincreases.

3. The strength of the interaction increases as the size of the colloidal particlesdecreases (at constant φc), because osmotic effects are influenced more by thenumber of particles involved (which increases with decreasing rc) than by thevolume of the depletion zone (which decreases with decreasing rc).

4. The range of the interaction increases as the size of the colloidal particles increases.

Equation 3.22 suggests that the strength of the depletion interaction is independent of pHand ionic strength. Nevertheless, these parameters may indirectly influence the depletioninteraction by altering the effective size of the colloidal particles and the depletion zone. Forexample, changing the number of charges on a biopolymer molecule by altering the pH caneither increase or decrease its effective size (Launay et al. 1986, Rha and Pradipasena 1986).Increasing the number of similarly charged groups usually causes a biopolymer to become

FIGURE 3.18 Influence of the volume fraction (φc) of colloidal particles on the attractive depletioninteraction between emulsion droplets (rd = 1 µm, rc = 10 nm).


more extended because of electrostatic repulsion between the groups. On the other hand,decreasing the number of similarly charged groups or having a mixture of positively andnegatively charged groups usually causes a biopolymer to reduce its effective size. Alteringthe ionic strength of an aqueous solution also causes changes in the effective size of biopoly-mer molecules (e.g., adding salt to a highly charged biopolymer molecule screens the elec-trostatic repulsion between charged groups and therefore causes a decrease in biopolymersize) (Launay et al. 1986, Rha and Pradipasena 1986). Thus, if the colloidal particles areionic, one would expect the strength of the depletion interaction to depend on pH and ionicstrength, but if they are nonionic, one would expect them to be fairly insensitive to theseparameters.

3.7. HYDROPHOBIC INTERACTIONS

Compared to the other major forms of colloidal interaction, the contribution of hydrophobicinteractions to emulsion stability has largely been ignored by emulsion scientists. Neverthe-less, this type of interaction is of great importance in many types of foods and has recentlybeen shown to promote droplet flocculation in protein-stabilized emulsions (Monahan et al.1996, Demetriades et al. 1997b). Hydrophobic interactions are important when the surfacesof the droplets have some nonpolar character, either because they are not completely coveredby emulsifier (e.g., during homogenization or at low emulsifier concentrations) or because theemulsifier has some hydrophobic regions exposed to the aqueous phase (e.g., adsorbedproteins). The origin of the hydrophobic interaction is the ability of water molecules to formrelatively strong hydrogens bonds with each other, but not with nonpolar molecules (Chapter4). Consequently, the interaction between nonpolar substances and water is thermodynami-cally unfavorable, which means that a system will attempt to minimize the contact areabetween these substances by causing them to associate (Tanford 1980, Israelachvili 1992,Israelachvili and Wennerstrom 1996, Alaimo and Kumosinski 1997). This process manifestsitself as a relatively strong attractive force between hydrophobic substances dispersed inwater and is responsible for many important phenomena which occur in food emulsions, suchas protein conformation, micelle formation, adsorption of surfactants, and the low watersolubility of nonpolar compounds (Chapter 4).

One of the major reasons that hydrophobic interactions were ignored in the past is thatthere were no theories available to predict their magnitude and range. The complex nature oftheir origin, which depends on changes in the interactions and structural organization of alarge number of water molecules in the vicinity of the nonpolar groups, means that it isextremely difficult to develop mathematical theories from first principles (Israelachvili 1992,Paulaitis et al. 1996). Nevertheless, recent advances in the development of sensitive instru-ments for measuring the forces between macroscopic bodies have enabled researchers todevelop empirical equations to describe the magnitude and range of hydrophobic interactions(Israelachvili and Pashley 1984, Pashley et al. 1985, Claesson 1987, Claesson and Christenson1988, Rabinovich and Derjaguin 1988). These experiments have shown that the hydrophobicinteraction between nonpolar surfaces is relatively strong and long range and that it decaysexponentially with surface-to-surface separation. Considerable progress in understanding thenature of hydrophobic interactions has also been achieved using computer simulations (Paulaitiset al. 1996).

The interdroplet pair potential between two emulsion droplets with hydrophobic surfacesseparated by water is given by (Israelachvili and Pashley 1984):

w h r eih

hydrophobic( ) /= − −2 00π γ φλ λ (3.23)


FIGURE 3.19 An attractive hydrophobic interaction arises between emulsion droplets when theirsurfaces have some hydrophobic character, where φ is approximately equal to the fraction of the dropletsurface which is nonpolar.

where γi is the interfacial tension between the nonpolar groups and water (typically between10 to 50 mJ m–2 for food oils), φ is a parameter that varies between 0 and 1 which takes intoaccount the fact that only part of the droplet surface is hydrophobic, and λ0 is the decay lengthof the interaction (typically between 1 to 2 nm) (Israelachvili 1992). This equation indicatesthat the magnitude of the hydrophobic interaction increases as the surfaces become morehydrophobic (i.e., φ tends toward unity). Experiments have shown that for bare nonpolarsurfaces, the hydrophobic attraction is stronger than the van der Waals attraction up toseparations of 80 nm (Israelachvili 1992).

When hydrophobic surfaces are covered by amphiphilic molecules, such as small-moleculesurfactants or biopolymers, the hydrophobic interaction between them is effectively screenedand the overall attraction is mainly due to van der Waals interactions (Israelachvili 1992).Nevertheless, hydrophobic interactions are significant when the surface has some hydropho-bic character (e.g., if the surface is not completely saturated with emulsifier molecules, if itis bent to expose the oil molecules below [Israelachvili 1992], or if the emulsifier moleculeshave some hydrophobic regions exposed to the aqueous phase [Demetriades et al. 1997b]).Experiments have shown that the hydrophobic interaction is not directly proportional to thenumber of nonpolar groups at a surface, because the alteration in water structure imposed bynonpolar groups is disrupted by the presence of any neighboring polar groups (Israelachvili1992). Thus it is not possible to assume that φ is simply equal to the fraction of nonpolar sitesat a surface. As a consequence, it is difficult to accurately predict their magnitude from firstprinciples.

Hydrophobic interactions become increasingly strong as the temperature is raised(Israelachvili 1992). Thus, hydrophobic interactions between emulsion droplets become moreimportant at higher temperatures. Because the strength of hydrophobic interactions dependson the magnitude of the interfacial tension, any change in the properties of the solvent whichincreases the interfacial tension will increase the hydrophobic attraction. The addition ofsmall amounts of alcohol to the aqueous phase of an emulsion lowers γi and therefore reducesthe hydrophobic attraction between nonpolar groups. Electrolytes which alter the structuralarrangement of water molecules also influence the magnitude of the hydrophobic effect when


they are present at sufficiently high concentrations (Christenson et al. 1990). Structurebreakers tend to enhance hydrophobic interactions, whereas structure promoters tend toreduce them (Chapter 5). Variations in pH have little direct effect on the strength of hydro-phobic interactions, unless there are accompanying alterations in the structure of the water orthe interfacial tension (Israelachvili and Pashley 1984).

3.8. HYDRATION INTERACTIONS

Hydration interactions arise from the structuring of water molecules around dipolar and ionicgroups (in contrast to hydrophobic interactions, which arise from the structuring of wateraround nonpolar groups). Most food emulsifiers naturally have dipolar or ionic groups thatare hydrated (e.g., –OH, –COO–, and –NH3

+), and some are also capable of binding hydratedions (e.g., –COO– + Na+ → –COO– Na+). As two droplets approach each other, the bondsbetween the polar groups and the water molecules in their immediate vicinity must bedisrupted, which results in a repulsive interaction (Figure 3.20) (Besseling 1997). The mag-nitude and range of the hydration interaction therefore depend on the number and strength ofthe bonds formed between the polar groups and the water molecules: the greater the degreeof hydration, the more repulsive and long range the interaction (Israelachvili 1992). Just aswith hydrophobic interactions, it is difficult to develop theories from first principles todescribe this type of interaction because of the complex nature of its origin and its depen-dence on the specific type of ions and polar groups present. Nevertheless, experimentalmeasurements of the forces between two liquid surfaces have shown that hydration interac-tions are fairly short-range repulsive forces that decay exponentially with surface-to-surfaceseparation (Claesson 1987, Israelachvili 1992):

w h Ar e hhydration( ) /= −λ λ

00 (3.24)

where A is a constant which depends on the degree of hydration of the surface (typicallybetween 3 and 30 mJ m–2) and λ0 is the characteristic decay length of the interaction (typically

FIGURE 3.20 Short-range repulsive interactions arise between emulsion droplets when they comeinto close contact due to hydration, protrusion, and undulation of interfacial layers.


between 0.6 and 1.1 nm) (Israelachvili 1992). The greater the degree of hydration of a surfacegroup, the larger the values of A and λ0. In practice, it is often difficult to isolate thecontribution of the hydration forces from other short-range interactions that are associatedwith mobile interfacial layers at small separations (such as steric and thermal fluctuationinteractions) (Figure 3.20), and so there is still much controversy about their origin andnature. Nevertheless, it is widely accepted that they make an important contribution to theoverall interaction energy in many systems.

At high electrolyte concentrations, it is possible for ionic surface groups to specificallybind hydrated ions to their surfaces (Hunter 1986, 1989; Miklavic and Ninham 1990). Someof these ions have large amounts of water associated with them and can therefore providestrong repulsive hydration interactions. Specific binding depends on the radius and valencyof the ion involved, because these parameters determine the degree of ion hydration. Ions thathave small radii and high valencies tend to bind less strongly because they are surroundedby a relatively thick layer of tightly “bound” water molecules and some of these must beremoved before the ion can be adsorbed (Israelachvili 1992). As a general rule, the adsorbabilityof ions from water can be described by a lyotropic series: I– > Br– > Cl– > F– for monovalentions, and K+ > Na+ > Li+ for monovalent cations (in order of decreasing adsorbability). Onthe other hand, once an ion is bound to a surface, the strength of the repulsive hydrationinteraction between the emulsion droplets increases with degree of ion hydration becausemore energy is needed to dehydrate the ion as the two droplets approach each other. There-fore, the ions that adsorb the least strongly are those that provide the greatest hydrationrepulsion. Thus it is possible to control the interaction between droplets by altering the typeand concentration of ions present in the aqueous phase.

Hydration interactions are often strong enough to prevent droplets from aggregating(Israelachvili 1992). Thus, oil-in-water emulsions that should contain enough electrolyte tocause droplet flocculation through electrostatic screening have been found to be stablebecause of specific binding of ions (Israelachvili 1992). This effect is dependent on the pHof the aqueous phase because the electrolyte ions have to compete with the H+ or OH– ionsin the water (Miklavic and Ninham 1990). For example, at relatively high pH and electrolyteconcentrations (>10 mM), it has been observed that Na+ ions can adsorb to negatively chargedsurface groups and prevent droplets from aggregating through hydration repulsion, but whenthe pH of the solution is decreased, the droplets aggregate because the high concentration ofH+ ions displaces the Na+ ions from the droplet surface (Israelachvili 1992). Nonionicemulsifiers are less sensitive to pH and ionic strength, and they do not usually bind highlyhydrated ions. The magnitude of the hydration interaction decreases with increasing tempera-ture because polar groups become progressively dehydrated as the temperature is raised(Israelachvili 1992). In summary, the importance of hydration interactions in a particularsystem depends on the nature of the hydrophilic groups on the droplet surfaces, as well as onthe type and concentration of ions present in the aqueous phase.

3.9. THERMAL FLUCTUATION INTERACTIONS

The interfacial region which separates the oil and aqueous phases of an emulsion is oftenhighly dynamic (Israelachvili 1992). In particular, interfaces that are comprised of small-molecule surfactants tend to exhibit undulations because their bending energy is relativelysmall compared to the thermal energy of the system (Figure 3.21a). In addition, the surfactantmolecules may be continually twisting and turning, as well as moving in and out of theinterfacial region (Figure 3.21b). When two dynamic interfaces move close to each other,they experience a number of repulsive thermal fluctuation interactions which are entropic in


(a) (b)

origin (Figure 3.20) (Israelachvili 1992). In emulsions, the two most important of these areprotrusion and undulation interactions.

3.9.1. Protrusion Interactions

Protrusion interactions are short-range repulsive interactions which arise when two surfacesare brought so close together that the movement of the surfactant molecules in and out of theinterface of one droplet is restricted by the presence of another droplet, which is entropicallyunfavorable. The magnitude of this repulsive interaction depends on the distance that thesurfactant molecules are able to protrude from the interface, which is governed by theirmolecular structure. The interdroplet pair potential due to protrusion interactions is given bythe following expression (Israelachvili 1992):

w h rkT e hprotrusion ( ) /≈ −3 0

0π λ λΓ (3.25)

where Γ is the number of surfactant molecules (or head groups) per unit surface area and λ0

is the characteristic decay length of the interaction (typically between 0.07 to 0.6 nm), whichdepends on the distance the surfactant can protrude from the surface.

3.9.2. Undulation Interactions

Undulation interactions are short-range repulsive interactions which arise when the wave-likeundulations of the interfacial region surrounding one emulsion droplet are restricted by thepresence of another emulsion droplet, which is entropically unfavorable. The magnitude andrange of this repulsive interaction increase as the amplitude of the oscillations increases. Theinterdroplet pair potential due to undulation interactions is given by the following expression(Israelachvili 1992):

w hr kT

k hbundulation ( )

( )≈ π 2

4 (3.26)

where kb is the bending modulus of the interfacial layer, which typically has values ofbetween about 0.2 and 20 × 10–20 J depending on the surfactant type. The magnitude of thebending modulus is related to the molecular geometry of the surfactant molecules (Chapter

FIGURE 3.21 Interfaces that are comprised of small-molecule surfactants are susceptible to protru-sion and undulation interactions.


4) and tends to be higher for surfactants that have two nonpolar chains than those that haveonly one.

Thermal fluctuation interactions are much more important for small-molecule surfactantswhich form flexible interfacial layers than for biopolymers which form rigid interfaciallayers. Both types of interaction tend to increase with temperature because the interfacesbecome more mobile. Nevertheless, this effect may be counteracted by increasing dehydra-tion of any polar groups with increasing temperature. The fact that these interactions areextremely short range means that they are unlikely to make any significant contribution to theflocculation of emulsion droplets. Nevertheless, they may be important in preventing dropletcoalescence because of the relatively large repulsive interaction which arises at close dropletseparations. The strength of this interaction is governed by the structure and dynamics of theinterfacial layer and therefore varies considerably from system to system (Israelachvili 1992).

3.10. HYDRODYNAMIC INTERACTIONS ANDNONEQUILIBRIUM EFFECTS

So far, it has been assumed that the interactions between droplets occur under equilibriumconditions. In practice, the droplets in an emulsion are in continual motion, which influencestheir interactions in a number of ways (Evans and Wennerstrom 1994). First, the system maynot have time to reach equilibrium when two droplets rapidly approach each other becausemolecular rearrangements (such as adsorption–desorption of emulsifiers, ionization/deioniza-tion of charged groups, and conformational changes of biopolymers) take a finite time tooccur (Israelachvili 1992, Israelachvili and Berman 1995). As a consequence, the colloidalinteractions between droplets may be significantly different from those observed underequilibrium conditions. These nonequilibrium effects depend on the precise nature of thesystem and are therefore difficult to account for theoretically. Second, the movement of adroplet causes an alteration in the flow profile of the surrounding liquid, which can be “felt”by another droplet (Dukhin and Sjoblom 1996). As two droplets move closer together, thecontinuous phase must be squeezed out from the narrow gap separating them against thefriction of the droplet surfaces. This effect manifests itself as a decrease in the effectivediffusion coefficient of the emulsion droplets, D(s) = D0G(s), where D0 is the diffusioncoefficient of a single droplet and G(s) is a correction factor which depends on the dimen-sionless separation between the droplets: s = (h + 2r)/r (Davis et al. 1989). Mathematicalexpressions for G(s) have been derived from a consideration of the forces that act on particlesas they approach each other in a viscous liquid (Davis et al. 1989, Zhang and Davis 1991,Dukhin and Sjoblom 1996). For rigid spherical particles, the hydrodynamic correction factorcan be approximated by the following expression (Dickinson and Stainsby 1982):

G sd d

d d( )

ƒ ƒ

ƒ ƒ= +

+ +6 4

6 13 2

2

2 2(3.27)

where d = (s – 2). The value of G(s) varies from 0 when the particles are in close contact (h= 0, s = 2) to unity when they are far apart and therefore have no influence on each other (h,s→ ∞). Thus, as particles approach each other, their speed gets progressively slower, andtherefore they would not aggregate unless there was a sufficiently strong attractive colloidalinteraction to overcome the repulsive hydrodynamic interaction. Equation 3.27 must bemodified for emulsions to take into account the fact that there is less resistance to themovement of the continuous phase out of the gap between the droplets when their surfaces


have some fluid-like characteristics (Davis et al. 1989, Zhang and Davis 1991). Thus thehydrodynamic resistance to the approach of fluid droplets is less than that for solid droplets.Hydrodynamic interactions are particularly important for determining the stability of dropletsto flocculation and coalescence in emulsion systems (Chapter 7).

3.11. TOTAL INTERACTION POTENTIAL

The overall interdroplet pair potential is the sum of the various attractive and repulsivecontributions:*

w h w h w h w h w h

w h w h w h

total VDW electrostatic steric depletion

hydrophobic hydration thermal

( ) ( ) ( ) ( ) ( )

( ) ( ) ( )

= + + +

+ + + (3.28)

Not all of these interactions play an important role in every type of food emulsion, and itis often possible to identify two or three interactions which dominate the overall interaction.For this reason, it is informative to examine the characteristics of certain combinations ofcolloidal interaction which are particularly important in food emulsions. A summary of thecharacteristics of the various types of interaction is given in Table 3.3.

3.11.1. van der Waals and Electrostatic

The classical approach to describing the interactions between charge-stabilized emulsiondroplets is the DLVO theory, which is named after the four scientists who first proposedit: Derjaguin, Landau, Verwey, and Overbeek (Hiemenz 1986, Hunter 1986, Derjaguin etal. 1987, Derjaguin 1989). This theory assumes that the overall interaction between a pairof emulsion droplets is the result of van der Waals attractive and electrostatic repulsiveinteractions:

w h w h w h( ) ( ) ( )= +VDW electrostatic (3.29)

The dependence of the van der Waals and electrostatic interactions on separation for twoelectrically charged oil droplets dispersed in water is illustrated in Figure 3.22. The van derWaals interaction potential is negative (attractive) at all separations, whereas the electro-static interaction potential is positive (repulsive). The overall interdroplet pair potential hasa more complex dependence on separation because it is the sum of two opposing interac-tions, and it may therefore be attractive at some separations and repulsive at others. Whenthe two droplets are separated by a large distance, there is no effective interaction betweenthem. As they move closer together, the van der Waals attraction dominates initially andthere is a shallow minimum in the profile, which is referred to as the secondary minimum,w(h2

omin). When the depth of this minimum is large compared to the thermal energy, w(h2

omin)

>> kT, the droplets tend to be flocculated, but if it is small compared to the thermal energy,the droplets tend to remain unaggregated. At closer separations, the repulsive electrostaticinteraction dominates and there is an energy barrier, w(hmax), which must be overcomebefore the droplets can come any closer. If this energy barrier is sufficiently large comparedto the thermal energy, w(hmax) >> kT, it will prevent the droplets from falling into the deep

* In reality, it is not always appropriate to simply sum the contribution from all of the separate interactionsbecause some of them are coupled (Ninham and Yaminsky 1997). Nevertheless, this approach gives a good firstapproximation.


TABLE 3.3Summary of the Characteristics of the Various Types of Colloidal InteractionBetween Emulsion Droplets

Affected Effect of Effect ofInteraction Sign Strength Range by pH ionic strength temperature

van der Waals A S LR No Reduced DecreasesElectrostatic R W → S SR → LR Yes Reduced IncreasesSteric

Mixing A or R W → S SR SD SD SDElastic R S SR SD SD Increases

Depletion A W → S SR SD SD IncreasesHydrophobic A S LR No Yes IncreasesHydration R S SR → LR Indirectly Indirectly DecreasesThermal R S SR Indirectly No Increases

fluctuation

Note: A = attractive, R = repulsive, S = strong, W = weak, SR = short range (<10 nm), LR = long range (>10 nm),and SD = system dependent.

primary minimum that exists at closer separations. On the other hand, if it is not largecompared to the thermal energy, then the droplets tend to fall into the primary minimum,w(h1

omin), which would lead to droplet coalescence. An example of this type of system would

be two electrically charged oil droplets suspended in pure water in the absence of anyadsorbed emulsifier.

Electrostatically stabilized emulsions are particularly sensitive to the ionic strength and pHof the aqueous phase (Figure 3.23). At low electrolyte concentrations, there may be asufficiently high energy barrier to prevent the droplets from coming close enough togetherto aggregate into the primary minimum (Figure 3.23). As the ion concentration is increased,the screening of the electrostatic interaction becomes more effective (Section 3.4.3), whichreduces the height of the energy barrier. Above a certain electrolyte concentration, often

FIGURE 3.22 Typical interdroplet pair potential for an electrostatically stabilized emulsion, wherethe only interactions are electrostatic repulsion and van der Waals attractions (i.e., DLVO theory).


referred to as the critical aggregation concentration or CAC, the energy barrier is no longerhigh enough to prevent the droplets from falling into the deep primary minimum, and so thedroplets tend to aggregate. This accounts for the susceptibility of many electrostaticallystabilized food emulsions to droplet aggregation when salt is added to the aqueous phase(Hunt and Dalgleish 1994, 1995; Demetriades et al. 1997a). The electrical charge of manyfood emulsifiers is sensitive to the pH of the aqueous phase. For example, the droplet chargeof protein-stabilized emulsions decreases as the pH tends toward the isoelectric point of theproteins, which reduces the magnitude of the electrostatic repulsion between the droplets.This accounts for the tendency of protein-stabilized emulsions to become aggregated whentheir pH is adjusted to the protein isoelectric point (Demetriades et al. 1997a).

3.11.2. van der Waals, Electrostatic, and Steric

One of the major limitations of the DLVO theory is that it does not take into account theshort-range repulsive interactions which arise between droplets when they come close to-gether (i.e., polymeric steric, hydration, and thermal fluctuation) (Israelachvili 1992, Evansand Wennerstrom 1994). Consequently, it predicts that two liquid droplets would coalesce atclose separations because of the large attraction between them. In practice, many of theemulsifiers used in food emulsions are capable of stabilizing the droplets through a combi-nation of electrostatic and short-range repulsive interactions. The DLVO theory can thereforebe extended by including the effects of these short-range repulsive interactions. For conve-nience, we will assume that only the polymeric steric repulsion needs to be included:

w h w h w h w h( ) ( ) ( ) ( )= + +VDW electrostatic steric (3.30)

The dependence of the interdroplet pair potential on droplet separation for the same systemas that shown in Figure 3.22 but with an additional polymeric steric contribution is shown inFigure 3.24. The major difference between this system and the one with no steric interactionsis that the droplets cannot get close enough together to coalesce. If the droplets do havesufficient energy to overcome the energy barrier, they fall into a deep primary minimum butcannot get any closer because of the extremely strong short-range repulsive interactions. In

FIGURE 3.23 The stability of emulsion droplets to aggregation decreases as the ionic strength of theintervening medium increases because of electrostatic screening effects.


FIGURE 3.24 The presence of an adsorbed layer of emulsifier molecules modifies the interactionbetween electrostatically stabilized droplets (r = 1 µm, I = 0.1 M). As the thickness of the interfaciallayer increases, the depth of the energy minimum decreases and it moves farther away from the dropletsurfaces.

this case, the droplets would tend to form strong flocs, rather than become coalesced. Itshould be noted that the adsorbed emulsifier layer would also influence the magnitude of thevan der Waals interaction by an amount that depends on its thickness and composition(Section 3.3.6). Inclusion of the polymeric steric interactions in the DLVO theory makes itmuch more realistic because most food emulsifiers are polymeric.

A practical example of a real system which approximates this type of behavior would bea protein-stabilized oil-in-water emulsion (Demetriades et al. 1997a). The droplets are stableto aggregation when they are highly charged (i.e., when the pH is significantly lower orhigher than the isoelectric point) and the Debye length is sufficiently long (i.e., low ionicstrengths). When the pH is adjusted close to the isoelectric point of the proteins or the ionicstrength is increased, the droplets flocculate because the van der Waals attraction dominatesthe electrostatic repulsion, but they do not coalesce because of the strong steric repulsionwhich operates at short separations.

3.11.3. van der Waals and Steric

Many food emulsifiers have no electrical charge, but they are still capable of stabilizingemulsions against aggregation because of the various short-range repulsive interactions men-tioned earlier. Emulsifiers of this type include nonionic surfactants (such as Tweens andSpans) and some biopolymers (such as modified starch). For convenience, we will considera system in which only van der Waals attraction and polymeric steric interactions occur:

w h w h w h( ) ( ) ( )= +VDW steric (3.31)

The dependence of the overall interdroplet pair potential on droplet separation for emul-sions with interfacial layers of different thickness is shown in Figure 3.25. It is assumed that


the continuous phase is an indifferent quality solvent for the polymer, so that the mixingcontribution to the polymeric steric interaction is zero (Section 3.5.2.1). At wide separations,the overall interaction between the droplets is negligible. As the droplets move closer to-gether, the attractive van der Waals interaction begins to dominate and so there is a netattraction between the droplets. However, once the droplets get so close together that theirinterfacial layers overlap, then the repulsive polymeric steric interaction dominates and thereis a net repulsion between the droplets. At a particular separation, there is a minima in theinterdroplet pair potential, and this is the location where the droplets tend to reside. If thedepth of this minima is large compared to the thermal energy of the system, the dropletsremain aggregated; otherwise, they move apart. The depth and position of the minimumdepend on the thickness and properties of the interfacial layer surrounding the droplets. Asthe thickness of the adsorbed layer increases, the repulsive interaction between dropletsbecomes more significant at larger separations, and consequently the depth of the minimadecreases. If the interfacial layer is sufficiently thick, it may prevent the droplets fromaggregating altogether, because the depth of the minima is relatively small compared to thethermal energy. This accounts for the effectiveness of uncharged polysaccharides (whichform thick interfacial layers) in preventing droplet flocculation and coalescence, whereasuncharged protein molecules (which form thin interfacial layers) can prevent coalescence butnot flocculation. As mentioned in Section 3.3.6, the adsorbed layer may have an additionalinfluence on the overall interaction potential due to its modification of the van der Waalsinteractions.

3.11.4. van der Waals, Electrostatic, and Hydrophobic

The droplet surfaces in many food emulsions acquire some hydrophobic character duringtheir manufacture, storage, or consumption. A typical example is a whey-protein–stabilizedemulsion which is subjected to heat (Monahan et al. 1996, Demetriades et al. 1997b). Heating

FIGURE 3.25 An adsorbed layer of emulsifier molecules at the surface of the droplets causes a short-range repulsive interaction which prevents droplets from coalescing. As the thickness of the interfaciallayer increases, the depth of the energy minimum decreases and it moves farther away from the dropletsurfaces.


the emulsion above 65°C causes the protein molecules adsorbed to the oil–water interface topartially unfold and thus expose some of the nonpolar amino acids to the aqueous phase. Theoverall interdroplet pair potential for this type of system is given by:

w h w h w h w h( ) ( ) ( ) ( )= + +VDW electrostatic hydrophobic (3.32)

The dependence of the overall interdroplet pair potential on droplet separation for the samesystem as shown in Figure 3.22 but for droplets which have different degrees of surfacehydrophobicity is shown in Figure 3.26. As the hydrophobicity of the droplet surface in-creases, the hydrophobic attraction increases, which causes a decrease in the height of theenergy barrier. When the surface hydrophobicity is sufficiently large, the energy barrierbecomes so small that the droplets aggregate into the primary minimum. This accounts forthe experimental observation that whey-protein–stabilized emulsions become more suscep-tible to aggregation when they are heated above a temperature where the protein moleculesunfold (Demetriades et al. 1997b).

3.11.5. van der Waals, Electrostatic, and Depletion

Depletion interactions are important when the continuous phase of an emulsion contains asignificant concentration of small colloidal particles, such as surfactant micelles or nonadsorbingbiopolymers (Dickinson and McClements 1995, Jenkins and Snowden 1996). The interdropletpair potential for a system in which depletion interactions are important is given by:

w h w h w h w h( ) ( ) ( ) ( )= + +VDW electrostatic depletion (3.33)

The variation of the interdroplet pair potential with droplet separation for this type ofsystem is shown in Figure 3.27. At low concentrations of colloidal particles, the energybarrier is sufficiently large to prevent the droplets from falling into the primary minimum. As

FIGURE 3.26 Influence of hydrophobic interactions on the interdroplet pair potential of electrostati-cally stabilized emulsions (r = 1 µm, I = 0.1 M).


FIGURE 3.27 Influence of depletion interactions on the interdroplet pair potential of an electrostati-cally stabilized emulsion. As the volume fraction of colloidal particles increases, the depth of thesecondary minima increases and the height of the energy barrier decreases.

the concentration of colloidal particles is increased, the attraction between the dropletsincreases, and eventually the energy barrier is reduced to such a low level that the dropletscan fall into the primary minima and become coalesced or strongly flocculated. In addition,the depth of the secondary minimum increases, which will promote weak flocculation. Anumber of workers have shown that depletion interactions promote droplet flocculation inemulsions when the concentration of surfactant or biopolymer exceeds some critical concen-tration (Sperry 1982, Aronson 1992, McClements 1994, Dickinson et al. 1995, Jenkins andSnowden 1996).

3.12. PREDICTING COLLOIDAL INTERACTIONSIN FOOD EMULSIONS

In this chapter, we have examined the origin, magnitude, and range of the most importanttypes of attractive and repulsive interactions which can arise between emulsion droplets. Inprinciple, it is possible to predict the likelihood that the droplets in an emulsion will be in anaggregated or an unaggregated state using the theories given above. In practice, it is ex-tremely difficult to make quantitative predictions about the properties of food emulsions fora number of reasons. First, food emulsions contain a huge number of different emulsiondroplets (rather than just two) which interact with each other and with other componentswithin the system, and it is difficult to quantify the overall nature of these interactions(Dickinson 1992). Second, there is often a lack of information about the relevant physicalparameters needed to carry out the calculations (Hunter 1986). Third, certain simplifyingassumptions often have to be made in the theories in order to derive tractable expressions forthe interaction energies, and these are not always justified (Ninham and Yaminsky 1997).Fourth, food systems are not usually at thermodynamic equilibrium and so the above equa-


tions do not strictly apply (Israelachvili and Berman 1995). Fifth, covalent interactions areimportant in some systems, and these are not taken into account in the above analysis(Mohanan et al. 1996). Finally, food emulsions may be subjected to external forces whichaffect the interactions between the droplets (e.g., gravity, centrifugation, or mechanicalagitation). Despite the limitations described above, an understanding of the various types ofinteractions which act between droplets gives food scientists a powerful tool for interpretingdata and for systematically accounting for the effects of ingredient formulations and process-ing conditions on the properties of many food products. It is often possible to predict themajor factors which determine the stability of emulsions (albeit in a fairly qualitative fash-ion). Rather than trying to theoretically predict the overall interaction between emulsiondroplets, it is often easier to directly measure the forces which act between them. Over thepast 30 years or so, a number of experimental techniques have been developed which allowscientists to accurately measure the forces between macroscopic surfaces down to separationsof a fraction of a nanometer (Israelachvili 1992, Luckham and Costello 1993, Claesson et al.1996). The application of these techniques is greatly advancing our understanding of thisimportant area and leading to exciting new insights into the relative contributions of thevarious types of interaction mentioned above.


4 Emulsion Ingredients

4.1. INTRODUCTION

Food emulsions are compositionally complex materials which contain a wide variety ofdifferent chemical constituents (Friberg and Larsson 1997, Dickinson and Stainsby 1982,Dickinson 1992) (Table 4.1). The composition of an emulsion can be defined in a number ofdifferent ways: concentrations of specific atoms (e.g., H, C, O, N, Na, Mg, Cl), concentrationsof specific molecules (e.g., water, sucrose, amylose, β-lactoglobulin), concentrations ofgroups of molecules (e.g., proteins, lipids, carbohydrates, minerals), and concentrations ofspecific ingredients (e.g., flour, milk, salt, egg). Food manufacturers are usually concernedwith the concentration of specific ingredients, because food components are usually pur-chased and utilized in this form. On the other hand, research scientists may be more interestedin the concentrations of specific atoms, molecules, or groups of molecules, depending on thepurpose of their investigations.

The constituents in a food emulsion interact with each other, either physically or chemi-cally, to determine the overall physicochemical and organoleptic properties of the finalproduct. The efficient production of high-quality food emulsions therefore depends on aknowledge of the contribution that each individual constituent makes to the overall propertiesand how this contribution is influenced by the presence of the other constituents. Selectionof the most appropriate constituents is one of the most important decisions that a foodmanufacturer must make during the formulation and production of a food product. Ingredi-ents must exhibit the desired functional properties while being of a reliably high standard, lowcost, and ready availability.

The formulation of food products has traditionally been more of a craft than a science.Many of the foods which are familiar to us today are the result of a long and complex historyof development. Consequently, there has often been a rather poor understanding of the role(or multiple roles) that each chemical constituent plays in determining the overall quality.This century has seen the development of large-scale industrial manufacturing operationswhere foods are mass produced. Mass production has led to the availability of a wide varietyof low-cost foods, which are quick and easy to prepare and are therefore appealing to themodern consumer. Nevertheless, increasing reliance on mass production has meant that foodmanufacturers have had to develop a more thorough understanding of the behavior of foodingredients before, during, and after processing (Hollingsworth 1995). This knowledge isrequired for a number of reasons:

1. The properties of the ingredients entering a food factory often vary from batch tobatch. Food manufacturers utilize their knowledge of the behavior of food ingre-dients under different conditions to adjust the food-processing operations so thatthe final product has consistent properties.

2. There is a growing trend toward improving the quality, variety, and convenienceof processed foods (Sloan 1994, 1996). Knowledge of ingredient properties en-


TABLE 4.1Ingredients Typically Found in Food Emulsions

Macrocomponents Microcomponents

Water EmulsifiersLipids MineralsProteins GumsCarbohydrates Flavors

ColorsPreservativesVitamins

ables food scientists to develop these foods in a more systematic and informedmanner.

3. There is an increasing tendency toward removing or reducing the amounts of foodconstituents which have been associated with human health concerns, such as fat,salt, and cholesterol (Sloan 1994, 1996). The removal of certain ingredients causessignificant changes in food properties which are perceived as being undesirable toconsumers (e.g., many no-fat or low-fat products do not exhibit the desirable tasteor textural characteristics of the full-fat products they are designed to replace)(O’Donnell 1995). Consequently, it is important to understand the role that eachingredient plays in determining the overall physicochemical and organolepticproperties of foods, so that this role can be mimicked by a healthier alternativeingredient.

4.2. FATS AND OILS

Fats and oils are part of a group of compounds known as lipids (Gunstone and Norris 1983,Weiss 1983, Nawar 1996). By definition, a lipid is a compound which is soluble in organicsolvents but insoluble or only sparingly soluble in water (Nawar 1996). This group ofcompounds contains a large number of different types of molecules, including acylglycerols,fatty acids, and phospholipids (Nawar 1996). Triacylglycerols are by far the most commonlipid in foods, and it is this type of molecule that is usually referred to as a fat or oil. Ediblefats and oils come from a variety of different sources including plants, seeds, nuts, animals,and fish (Sonntag 1979c, Weiss 1983, Nawar 1996). By convention a fat is solid-like at roomtemperature, whereas an oil is liquid, although these terms are often used interchangeably(Walstra 1987). Because of their high natural abundance and their great importance in foods,we will be concerned almost exclusively with the properties of triacylglycerols in this section.

Fats and oils influence the nutritional, organoleptic, and physicochemical properties offood emulsions in a variety of ways. They are a major source of energy and essentialnutrients; however, overconsumption of certain types of these constituents leads to obesityand human health concerns (Chow 1992, Smolin and Grosvenor 1994). Consequently, therehas been a trend in the food industry to reduce the fat content of many traditional foods(O’Donnell 1995). The challenge for the food scientist is to create a product which has thesame desirable quality attributes as the original, but with a reduced fat content, which is oftenextremely difficult. The perceived flavor of a food emulsion is determined by the type andconcentration of lipids present (McNulty 1987, Overbosch et al. 1991, Landy et al. 1996).Lipids undergo a variety of chemical changes during the processing, storage, and handling


of foods that generate products which can be either desirable or deleterious to their flavorprofile (Nawar 1996). Controlling these reactions requires a knowledge of both lipid chem-istry and emulsion science (Frankel 1991, Frankel et al. 1994, Coupland and McClements1996). The taste and aroma of food emulsions are also governed by the partitioning of flavorcompounds between the oil, water, interfacial, and gaseous phases (Chapter 9). The oil phasemay also act as a solvent for various other important food components, including oil-solublevitamins, antioxidants, preservatives, and essential oils. Reducing the fat content of anemulsion can therefore have a profound influence on its flavor profile, stability, and nutri-tional content.

Food emulsions usually appear either cloudy or opaque because the light passing throughthem is scattered by the droplets (Hernandez et al. 1991). Emulsion appearance is thereforethe direct result of the immiscibility of oil and water. The characteristic texture of many foodemulsions is due to the ability of the oil phase to crystallize (Mulder and Walstra 1974,Walstra 1987, Moran 1994). The “spreadability” of margarines and butters is determined bythe formation of a three-dimensional network of aggregated fat crystals in the continuousphase which provides the product with mechanical rigidity (Moran 1994). The tendency ofa cream to thicken or “clot” when it is cooled below a certain temperature is due to theformation of fat crystals in the oil droplets, which causes them to aggregate (Boode 1992).The melting of fat crystals in the mouth causes a cooling sensation, which is an importantsensory attribute of many fatty foods (Walstra 1987). The ability of food scientists to improvethe quality of food emulsions therefore depends on an improved understanding of the multipleroles that fats and oils play in determining their properties.


Chemically, triacylglycerols are esters of a glycerol molecule and three fatty acid molecules(Figure 4.1). Each of the fatty acids may contain different numbers of carbon atoms and mayhave different degrees of unsaturation and branching (Lawson 1995, Nawar 1996). Neverthe-less, most naturally occurring fatty acids have an even number of carbon atoms (usually lessthan 24) and are straight chained. The fact that there are many different types of fatty acidmolecules, and that these fatty acids can be located at different positions on the glycerolmolecule, means that there are a huge number of possible triacylglycerol molecules presentin foods. Indeed, edible fats and oils always contain a great many different types oftriacylglycerol molecules, with the precise type and concentration depending on their origin(Weiss 1983).

Triacylglycerol molecules have a “tuning-fork” structure, with the two fatty acids at theends of the glycerol molecule pointing in one direction and the fatty acid in the middlepointing in the opposite direction (Figure 4.1). Triacylglycerols are predominantly nonpolarmolecules, and so the most important types of molecular interactions with their neighbors are

FIGURE 4.1 Chemical structure of a triacylglycerol molecule, which is assembled from three fattyacids and a glycerol molecule.


van der Waals attraction and steric overlap repulsion (Chapter 2). At a certain molecularseparation, there is a minimum in the intermolecular pair potential whose depth is a measureof the strength of the attractive interactions which bind the molecules together in the solid andliquid states (Section 2.4). Whether a triacylglycerol exists as a gas, liquid, or solid at aparticular temperature depends on a balance between these attractive interactions and thedisorganizing influence of the thermal energy (Section 4.2.3).

4.2.2. Bulk Physicochemical Properties

The bulk physicochemical properties of edible fats and oils depend on the molecular struc-ture and interactions of the triacylglycerol molecules they contain (Formo 1979; Gunstoneand Norris 1983; Birker and Padley 1987; Timms 1991, 1995). The strength of the attractiveinteractions between molecules and the effectiveness of their packing in a condensed phasedetermine their melting point, boiling point, density, and compressibility (Israelachvili1992). Triacylglycerols which contain branched or unsaturated fatty acids are not able topack as closely together as those which contain linear saturated fatty acids, and so they havelower melting points, lower densities, and higher compressibilities than saturatedtriacylglycerols (Walstra 1987). The melting points of triacylglycerols increases with in-creasing chain length, is higher for saturated than for unsaturated fatty acids, is higher forstraight-chain than branched fatty acids, and is higher for triacylglycerols with a moresymmetrical distribution of fatty acids on the glycerol molecule (Table 4.2). The relativelylow dielectric constant of triacylglycerol molecules compared to water is due to their smallpolarity (Table 4.3).

4.2.3. Fat Crystallization

One of the most important characteristics of fats and oils is their ability to undergo solid–liquid phase transitions at temperatures which occur during the processing, storage, andhandling of food emulsions (Birker and Padley 1987; Walstra 1987; Timms 1991, 1995). Thetexture, mouthfeel, stability, and appearance of many food emulsions depend on the physicalstate of the lipid phase (Moran 1994). The conversion of milk into butter relies on thecontrolled destabilization of an oil-in-water emulsion (milk) into a water-in-oil emulsion(butter), which is initiated by the formation of crystals in the milk fat globules (Mulder andWalstra 1974, Boode 1992). The spreadability of the butter produced by this process isgoverned by the final concentration of fat crystals (Moran and Rajah 1994). If the percentage

TABLE 4.2Melting Points of the Most Stable Polymorphof Selected Triacylglycerol Molecules

Triglyceride Melting point (°C)

LLL 46MMM 54PPP 64SSS 73OOO 5SOS 42SSO 38SOO 24


TABLE 4.3Comparison of the Bulk Physicochemical Propertiesof an Oil (Triolein) and Water at 20°C

Oil Water

Molecular weight 885 18Melting point (°C) 5 0Density (kg m–3) 910 998Compressibility (ms2 kg–1) 5.03 × 10–10 4.55 × 10–10

Viscosity (mPa s) ≈50 1.002Thermal conductivity (W m–1 K–1) 0.170 0.598Specific heat capacity (J kg–1 K–1) 1980 4182Thermal expansion coefficient (°C–1) 7.1 × 10–4 2.1 × 10–4

Dielectric constant 3 80.2Surface tension (mN m–1) ≈35 72.8Refractive index 1.46 1.333

FIGURE 4.2 The arrangement of triacylglycerols in the solid and liquid states depends on a balancebetween the organizing influence of the attractive interactions between the molecules and the disorga-nizing influence of the thermal energy.

of fat crystals is too high, the product is firm and difficult to spread, and if it is too low, theproduct is soft and tends to collapse under its own weight. The creation of food emulsionswith desirable properties therefore depends on an understanding of the major factors whichinfluence the crystallization and melting of lipids in foods (Birker and Padley 1987).

The arrangement of triacylglycerol molecules in the solid and liquid state is shown sche-matically in Figure 4.2. The physical state of a triacylglycerol at a particular temperaturedepends on its free energy, which is made up of contributions from enthalpic and entropicterms: ∆GS→L = ∆HS→L – T∆SS→L

(Atkins 1994). The enthalpy term (∆HS→L) represents thechange in the overall strength of the molecular interactions between the triacylglycerols whenthey are converted from a solid to a liquid, whereas the entropy term (∆SS→L) represents thechange in the organization of the molecules that is brought about by the melting process. Thestrength of the bonds between the molecules is greater in the solid than in the liquid statebecause the molecules are able to pack more efficiently, and so ∆HS→L is positive, whichfavors the solid state. On the other hand, the entropy of the molecules in the liquid state isgreater than that in the solid state, and therefore ∆SS→L is positive, which favors the liquidstate. At low temperatures, the enthalpy term dominates the entropy term (∆HS→L > T∆SS→L),and therefore the solid state has the lowest free energy (Atkins 1994). As the temperatureincreases, the entropic contribution becomes increasingly important. Above a certain tem-perature, known as the melting point, the entropy term dominates the enthalpy term (T∆SS→L

> ∆HS→L), and so the liquid state has the lowest free energy. A material therefore changesfrom a solid to a liquid when its temperature is raised above the melting point. This process


FIGURE 4.4 When there is a sufficiently high activation energy between the solid and liquid states,a liquid oil can persist in a metastable state below the melting point of a fat.

FIGURE 4.3 Temperature dependence of the free energies of the solid and liquid states. At lowtemperatures, the solid state is thermodynamically favorable, but above the melting point, the liquidstate is more favorable.

leads to the release of heat because of the reduction in the amount of energy stored in theintermolecular interactions when the material changes from a solid to a liquid (i.e., it is anexothermic process).

The temperature dependence of the free energies of the solid and liquid states clearlyshows that below the melting point, the solid state has the lowest free energy, but above it,the liquid state has the lowest (Figure 4.3). Thermodynamics informs us whether or not aphase transition can occur, but it tells us nothing about the rate at which this process occursor about the physical mechanism by which it is accomplished (Atkins 1994). As will be seenbelow, an understanding of lipid phase transitions requires a knowledge of both the thermo-dynamics and kinetics of the process. The crystallization of fats can be conveniently dividedinto three stages: supercooling, nucleation, and crystal formation (Boistelle 1988, Mullin1993).

4.2.3.1. Supercooling

Crystallization can only take place once a liquid phase is cooled below its melting point(Garside 1987, Walstra 1987). Even so, a material may persist as a liquid below its meltingpoint for a considerable time before any crystallization is observed (Skoda and van denTempel 1963, Phipps 1964, Mulder and Walstra 1974). This is because of an activationenergy which must be overcome before the liquid–solid phase transition can occur (Figure4.4). If the magnitude of this activation energy is sufficiently high compared to the thermalenergy of the system, crystallization will not occur, even though the transition is thermody-


namically favorable (Turnbull and Cormia 1961). The system is then said to exist in ametastable state. The height of the activation energy depends on the ability of crystal nucleiwhich are stable enough to grow into crystals to be formed in the liquid oil (see Section4.2.3.2.). The degree of supercooling of a liquid is defined as ∆T = T – Tmp, where T is thetemperature and Tmp is the melting point. The value of ∆T at which crystallization isobserved depends on the chemical structure of the oil, the presence of any contaminatingmaterials, the cooling rate, and the application of external forces (Dickinson and McClements1995). Pure oils containing no impurities can often be supercooled by more than 10°Cbefore any crystallization is observed (Turnbull and Cormia 1961; Dickinson et al. 1990, 1991c;McClements et al. 1993e).

4.2.3.2. Nucleation

Crystal growth can only occur after stable nuclei have been formed in a liquid. These nucleiare clusters of oil molecules that form small-ordered crystallites and are formed when anumber of oil molecules collide and become associated with each other (Hernqvist 1984).There is a free energy change associated with the formation of one of these nuclei (Garside1987). Below the melting point, the crystalline state is thermodynamically favorable, and sothere is a decrease in free energy when some of the oil molecules in the liquid cluster togetherto form a nucleus. This negative free energy (∆GV) change is proportional to the volume ofthe nucleus formed. On the other hand, the formation of a nucleus leads to the creation of anew interface between the solid and liquid phases which requires an input of energy toovercome the interfacial tension (Chapter 5). This positive free energy (∆GS) change isproportional to the surface area of the nucleus formed. The total free energy change associ-ated with the formation of a nucleus is therefore a combination of a volume and a surface term(Garside 1987):

∆ ∆ ∆∆ ∆

G G G rH T

TrV S i= + = +4

33 24π π γfus

mp(4.1)

where r is the radius of the nuclei, ∆Hfus is the enthalpy change per unit volume associatedwith the liquid–solid transition (which is negative), and γi is the solid–liquid interfacialtension. The volume contribution becomes increasingly negative as the size of the nucleiincreases, whereas the surface contribution becomes increasingly positive (Figure 4.5). Thesurface contribution dominates for small nuclei, while the volume term dominates for large

FIGURE 4.5 The critical size of a nucleus required for crystal growth depends on a balance betweenthe volume and surface contributions to the free energy of nuclei formation.


nuclei. At a certain critical nucleus radius (r*), the overall free energy has a maximum valuegiven by:

d Gdr

rH T

Tr i

∆ ∆ ∆= + =4 8 02π π γfus

mp(4.2)

This equation can be rearranged to give an expression for the critical radius of the nucleuswhich must be achieved for crystallization to occur:

rT

H Ti

* =2γ mp

fus∆ ∆ (4.3)

If a nucleus that has a radius below this critical size is formed, it will tend to dissociateso as to reduce the free energy of the system. On the other hand, if a nucleus that has a radiusabove this critical value is formed, it wil l tend to grow into a crystal. Equation 4.3indicates that the critical size of nuclei required for crystal growth decreases as the degreeof supercooling increases, which accounts for the increase in nucleation rate with decreasingtemperature.

The rate at which nucleation occurs can be related to the activation energy (∆G*), whichmust be overcome before a stable nucleus is formed (Boistelle 1988):

J A G kT= −exp( * / )∆ (4.4)

where J is the nucleation rate, which is equal to the number of stable nuclei formed per secondper unit volume of material; A is a preexponential factor; k is Boltzmann’s constant; and Tis the absolute temperature. The value of ∆G* is calculated by replacing r in Equation 4.1with the critical radius given in Equation 4.3. The variation of the nucleation rate predictedby Equation 4.4 with the degree of supercooling (∆T) is shown in Figure 4.6. The formationof stable nuclei is negligibly slow at temperatures just below the melting point, but increasesdramatically when the liquid is cooled below a certain temperature (T*). In reality, thenucleation rate increases with cooling up to a certain temperature, but then decreases onfurther cooling. This is because the increase in viscosity of the oil which occurs as the

FIGURE 4.6 Theoretically, the rate of the formation of stable nuclei increases with supercooling(solid line), but in practice, the nucleation rate decreases below a particular temperature because thediffusion of oil molecules is retarded by the increase in oil viscosity (broken line).


temperature is decreased slows down the diffusion of oil molecules toward the liquid–nucleusinterface (Boistelle 1988). Consequently, there is a maximum in the nucleation rate at aparticular temperature (Figure 4.6).

The type of nucleation described above occurs when there are no impurities present in theoil and is usually referred to as homogeneous nucleation (Boistelle 1988). If impurities suchas dust particles are present in the oil, the internal surface of the vessel containing the oil, theinterface of an emulsion droplet, the surface of an air bubble, or monoglyceride reversemicelles, then nucleation can be induced at a higher temperature than expected for a puresystem (Walstra 1987, McClements et al. 1993e, Dickinson and McClements 1995). Nucle-ation in the presence of impurities is referred to as heterogeneous nucleation and can bedivided into two types: primary and secondary (Boistelle 1988). Primary heterogeneousnucleation occurs when the impurities have a different chemical structure to that of the oil,whereas secondary heterogeneous nucleation occurs when crystals of the same chemicalstructure are present in the oil. Heterogeneous nucleation occurs when the impurities providea surface where the formation of stable nuclei is more energetically favorable than in the pureoil (Boistelle 1988). As a result, the degree of supercooling required to initiate dropletcrystallization is reduced. On the other hand, certain types of impurities are capable ofdecreasing the nucleation rate of oils because they are incorporated into the surface of thegrowing nuclei and therefore prevent any further oil molecules from being incorporated.Whether an impurity acts as a catalyst or an inhibitor of crystal growth depends on itsmolecular structure and interactions with the nuclei (Boistelle 1988).

4.2.3.3. Crystal Growth

Once stable nuclei have formed, they grow into crystals by incorporating molecules fromthe liquid oil at the solid–liquid interface (Garside 1987, Boistelle 1988). The crystal growthrate depends on the diffusion of a molecule from the liquid to the solid–liquid interface, aswell as its incorporation into the crystal lattice. Either of these processes may be ratelimiting, depending on the temperature and the molecular properties of the system. Moleculesmust encounter the interface at a location where crystal growth is favorable and also havethe appropriate shape, size, and orientation to be incorporated (Timms 1991, 1995). At hightemperatures, the incorporation of a molecule at the crystal surface is usually rate limiting,whereas at low temperatures, the diffusion step is rate limiting. This is because the viscosityof the liquid oil increases as the temperature is lowered, and so the diffusion of a moleculeis retarded. The crystal growth rate therefore increases initially with supercooling, has amaximum rate at a certain temperature, and then decreases on further supercooling (i.e., itshows a similar trend to the nucleation rate) (Figure 4.6). Nevertheless, the maximum rate ofnuclei formation may occur at a different temperature than the maximum rate of crystalgrowth. Experimentally, it has been observed that the rate of crystal growth is proportionalto the degree of supercooling and inversely proportional to the viscosity of the melt (Timms1991).

The crystal growth rate of fats can often be described by the following expression (Ozilgenet al. 1993):

1 1

0C t Ck tG( )

= + (4.5)

where C0 is the initial fraction of unsolidified oil, C(t) is the fraction at time t, and kG is thecrystal growth rate. Thus, a plot of 1/C versus time should give a straight line with a slope


equal to kG. One of the major problems in determining kG is that the nucleation rate and thecrystal growth rate often have similar magnitudes, and thus it is difficult to unambiguouslyestablish the contribution of each process (Ozilgen et al. 1993).

4.2.3.4. Crystal Morphology

The morphology of the crystals formed depends on a number of internal (e.g., packing ofmolecules and intermolecular forces) and external (e.g., temperature, cooling rate, mechani-cal agitation, and impurities) factors. When an oil is cooled rapidly to a temperature wellbelow its melting point, a large number of small crystals are formed, but when it is cooledslowly to a temperature just below its melting point, a smaller number of larger crystals areformed (Moran 1994, Timms 1995). This is because the nucleation rate increases morerapidly with decreasing temperature than the crystallization rate (Timms 1991). Thus, rapidcooling produces many nuclei simultaneously which subsequently grow into small crystals,whereas slow cooling produces a smaller number of nuclei which have time to grow intolarger crystals before further nuclei are formed. Crystal size has important implications forthe rheology and organoleptic properties of foods. When crystals are too large, they areperceived as being “grainy” or “sandy” in the mouth (Walstra 1987). The efficiency ofmolecular packing in crystals also depends on the cooling rate. If a fat is cooled slowly, orthe degree of supercooling is small, then the molecules have sufficient time to be efficientlyincorporated into a crystal (Walstra 1987). At faster cooling rates, or high degrees of super-cooling, the molecules do not have sufficient time to pack efficiently before another moleculeis incorporated. Thus rapid cooling tends to produce crystals which contain more dislocationsand in which the molecules are less densely packed (Timms 1991).

4.2.3.5. Polymorphism

Triacylglycerols exhibit a phenomenon known as polymorphism, which is the ability of amaterial to exist in a number of crystalline structures with different molecular packing(Hauser 1975, Garti and Sato 1988, Sato 1988, Hernqvist 1990). The three most commonlyoccurring types of packing in triacylglycerols are hexagonal, orthorhombic, and triclinic,which are usually designated as α, β′, and β polymorphic forms (Nawar 1996). The thermo-dynamic stability of the three forms decreases in the order β > β′ > α. Even though the β formis the most thermodynamically stable, triacylglycerols often crystallize in one of the meta-stable states because they have a lower activation energy of nuclei formation (Figure 4.7).With time, the crystals transform to the most stable state at a rate which depends on environ-mental conditions such as temperature, pressure, and the presence of impurities (Timms1991).

FIGURE 4.7 The polymorphic state that is initially formed when an oil crystallizes depends on therelative magnitude of the activation energies associated with nuclei formation.


4.2.3.6. Crystallization of Edible Fats and Oils

The melting point of a triacylglycerol depends on the chain length, branching, and degree ofunsaturation of its constituent fatty acids, as well as their relative positions along the glycerolmolecule (Table 4.3). Edible fats and oils contain a complex mixture of many different typesof triacylglycerol molecules, each with a different melting point, and so they melt over a widerange of temperatures rather than at a distinct temperature, as would be the case for a puretriacylglycerol (Figure 4.8).

The melting profile of a fat is not simply the weighted sum of the melting profiles of itsconstituent triacylglycerols, because high-melting-point triacylglycerols are soluble in lowermelting point ones (Timms 1995). For example, in a 50:50 mixture of tristearin and triolein,it is possible to dissolve 10% of solid tristearin in liquid triolein at 60°C (Walstra 1987,Timms 1995). The solubility of a solid component in a liquid component can be predicted,assuming they have widely differing melting points (>20°C):

ln xH

R T T= −

∆ fus

mp

1 1(4.6)

Here, x is the solubility, expressed as a mole fraction, of the higher melting point componentin the lower melting point component, and ∆Hfus is the molar heat of fusion (Walstra 1987).The structure and physical properties of crystals produced by cooling a complex mixture oftriacylglycerols are strongly influenced by the cooling rate and temperature (Moran 1994). Ifan oil is cooled rapidly, all the triacylglycerols crystallize at approximately the same time anda solid solution is formed, which consists of homogeneous crystals in which the triacylglycerolsare intimately mixed with each other (Walstra 1987). On the other hand, if the oil is cooledslowly, the higher melting point triacylglycerols crystallize first, while the low-melting-pointtriacylglycerols crystallize later, and so mixed crystals are formed. These crystals are hetero-geneous and consist of some regions which are rich in high-melting-point triacylglycerols andother regions that are depleted in these triacylglycerols. Whether a crystalline fat forms mixedcrystals or a solid solution influences many of its physicochemical properties, such as density,compressibility, and melting profile (Walstra 1987).

FIGURE 4.8 Comparison of the melting profile of a pure triacylglycerol and a typical edible fat. Theedible fat melts over a much wider range of temperatures because it consists of a mixture of manydifferent pure triacylglycerol molecules, each with different melting points.


Once a fat has crystallized, the individual crystals may aggregate to form a three-dimen-sional network which traps liquid oil through capillary forces (Moran 1994). The interactionsresponsible for crystal aggregation in pure fats are primarily van der Waals interactionsbetween the solid fat crystals, although “water bridges” between the crystals have also beensuggested to play an important role (Dickinson and McClements 1995). Once aggregation hasoccurred, the fat crystals may partially fuse together, which strengthens the crystal network(Timms 1995). The system may also change over time due to the growth of larger crystalsat the expense of smaller ones (i.e., Ostwald ripening) (Chapter 7).

4.2.3.7. Crystallization in Oil-in-Water and Water-in-Oil Emulsions

The influence of fat crystallization on the bulk physicochemical properties of food emulsionsdepends on whether the fat forms the continuous phase or the dispersed phase. The charac-teristic stability and rheological properties of water-in-oil emulsions, such as butter andmargarine, are determined by the presence of a fat crystal network in the continuous phase(Moran 1994, Chyrsam 1996). The fat crystal network is responsible for preventing the waterdroplets from sedimenting under the influence of gravity, as well as determining thespreadability of the product. If there are too many fat crystals present, the product is firm anddifficult to spread, but when there are too few crystals present, the product is soft andcollapses under its own weight. Selection of a fat with the appropriate melting characteristicsis therefore one of the most important aspects of margarine and spread production (Gunstoneand Norris 1983). The melting profile of natural fats can be optimized for specific applica-tions by various physical or chemical methods, including blending, interesterification, frac-tionation, and hydrogenation (Birker and Padley 1987).

Fat crystallization also has a pronounced influence on the physicochemical properties ofmany oil-in-water emulsions, such as milk or salad cream (Mulder and Walstra 1974, Boode1992). When the fat droplets are partially crystalline, a crystal from one droplet can penetrateanother droplet during a collision, which causes the two droplets to stick together (Walstra1987, Boode 1992, Dickinson and McClements 1995). This phenomenon is known as partialcoalescence and leads to a dramatic increase in the viscosity of an emulsion, as well as adecrease in the stability to creaming (Chapter 7). Extensive partial coalescence can eventuallylead to phase inversion (i.e., conversion of an oil-in-water emulsion to a water-in-oil emul-sion) (Mulder and Walstra 1974). This process is one of the most important steps in theproduction of butters, margarines, and spreads (Moran and Rajah 1994).

4.2.4. Chemical Changes

The two most important chemical changes which occur naturally in edible fats and oils arelipolysis and oxidation (Sonntag 1979b, Nawar 1996). Lipolysis is the process where esterbonds of fats and oils are hydrolyzed by certain enzymes or by a combination of heat andmoisture. The result of lipolysis is the liberation of free fatty acids, which can be eitherdetrimental or desirable to food quality. Lipolysis has deleterious effects on the quality ofsome food products because it leads to the generation of rancid off-flavors and off-odors(“hydrolytic rancidity”). In addition, free fatty acids are more surface active than triacylglycerolsand therefore accumulate preferentially at an oil–water or air–water interface, which in-creases their susceptibility to oxidation and may increase the tendency for emulsion dropletsto coalesce (Coupland and McClements 1996). On the other hand, a limited amount oflipolysis is beneficial to the quality of some foods because it leads to the formation ofdesirable flavors and aromas (e.g., cheese and yogurt) (Nawar 1996).


Lipid oxidation is one of the most serious causes of quality deterioration in many foodsbecause it leads to the generation of undesirable off-flavors and off-odors (“oxidative rancid-ity”), as well as potentially toxic reaction products (Schultz et al. 1962, Simic et al. 1992,Nawar 1996). In other foods, a limited amount of lipid oxidation is beneficial because it leadsto the generation of a desirable flavor profile (e.g., cheese). The term lipid oxidation describesan extremely complex series of chemical reactions which involves unsaturated lipids andoxygen (Halliwell and Gutterridge 1991, Nawar 1996). It has proved convenient to dividethese reactions into three different types: initiation, propagation, and termination (Halliwelland Gutterridge 1991). Initiation occurs when a hydrogen atom is extracted from the meth-ylene group (–CH=CH–) of a polyunsaturated fatty acid, leading to the formation of a freeradical (–CH=C–). This process can be started by a variety of different initiators that arepresent in foods, including naturally occurring lipid peroxides, transition metal ions, UVlight, and enzymes (Nawar 1996). It is worthwhile noting that many of these initiators arepredominantly water soluble, which has important implications for the oxidation of emulsi-fied oils, because the initiator must either travel through or interact across the interfacialmembrane in order to come into contact with the oil (Coupland and McClements 1996). Oncea free radical has formed, it reacts with oxygen to form a peroxy radical (–CH–COO–). Theseradicals are highly reactive and can extract hydrogen atoms from other unsaturated lipids andtherefore propagate the oxidation reaction. Termination occurs when two radicals interactwith each other to form a nonradical and thus end their role as propagators of the reaction.During lipid oxidation, a number of decomposition reactions occur simultaneously whichleads to the formation of a complex mixture of reaction products, including aldehydes,ketones, alcohols, and hydrocarbons (Nawar 1996). Many of these products are volatile andtherefore contribute to the characteristic odor associated with lipid oxidation. Some of theproducts are surface active and accumulate at oil–water or air–water interfaces, whereasothers are water soluble and would therefore leach into the aqueous phase of an emulsion(Coupland and McClements 1996).

A considerable amount of fundamental research has been carried out by lipid chemists toestablish the mechanism of lipid oxidation in bulk fats and oils (Nawar 1996). On the otherhand, the understanding of lipid oxidation in emulsions is still in its infancy, and a great dealof research needs to be carried out to determine the relationship between emulsion structureand lipid oxidation (Frankel 1991, Frankel et al. 1994, Coupland and McClements 1996).Some of the work carried out in this area is discussed in the chapter on emulsion stability(Chapter 7).

4.3. WATER

Water plays an extremely important role in determining the bulk physicochemical andorganoleptic properties of food emulsions. Its unique molecular and structural propertieslargely determine the solubility, conformation, and interactions of the other ingredientspresent in aqueous solutions. It is therefore important for food scientists to understand thecontribution that water makes to the overall properties of food emulsions.


A water molecule is comprised of two hydrogen atoms covalently bonded to an oxygen atom(Figure 4.9). The oxygen atom is highly electronegative and pulls the electrons associatedwith the hydrogen atoms toward it (Eisenberg and Kauzmann 1969, Kern and Karplus 1972,Fennema 1996b). This leaves a partial positive charge (δ+) on each of the hydrogen atoms


FIGURE 4.9 Molecular structure and tetrahedral organization of water molecules.

and a partial negative charge (δ–) on each of the lone pairs of electrons on the oxygen atom.The tetrahedral arrangement of the partial charges on a water molecule means that it canform hydrogen bonds with four of its neighbors (Figure 4.9). A hydrogen bond is formedbetween a lone pair of electrons on the oxygen atom of one water molecule and a hydrogenatom on a neighboring water molecule (i.e., O–Hδ+... Oδ–). A hydrogen bond is actually acomposite of more fundamental interactions (i.e., dipole–dipole, van der Waals, steric, andpartial charge transfer) (Baker and Hubbard 1984, Dill 1990). The magnitude of the hydro-gen bonds in water is typically between 13 and 25 kJ mol–1 (5 to 10 kT), which is sufficientlystrong to cause the water molecules to overcome the disorganizing influence of the thermalenergy and become aligned with each other (Israelachvili 1992). In order to maximize thenumber of hydrogen bonds formed, water molecules organize themselves into a three-dimensional tetrahedral structure because this allows each water molecule to form hydrogenbonds with four of its nearest neighbors (Franks 1972–82, Fennema 1996b). In the solidstate, the number of hydrogen bonds formed per molecule is four. In the liquid state, thedisorganizing influence of the thermal energy means that the number of hydrogen bonds permolecule is between about 3 and 3.5 at room temperature and decreases with increasingtemperature (Israelachvili 1992). The three-dimensional tetrahedral structure of water in theliquid state is highly dynamic, with hydrogen bonds continually being broken and reformedas the water molecules move about (Reichardt 1988, Franks 1991, Cramer and Truhlar 1994,Robinson et al. 1996). Water molecules which dissociate to form ions, such as H3O+ andOH–, do not fit into the normal tetrahedral structure of water; nevertheless, they have littleeffect on the overall structure and properties of water because their concentration is so low(Fennema 1996b).

In addition to forming hydrogen bonds with each other, water molecules are also capableof forming them with other polar molecules, such as organic acids, bases, proteins, andcarbohydrates (Franks 1972–82). The strength of these interactions varies from about 2 to 40kJ mol–1 (1 to 16 kT) depending on the electronegativity and orientation of the donor oracceptor groups (Baker and Hubbard 1984). Many ions form relatively strong ion–dipoleinteractions with water molecules, which has a pronounced influence on the structure andphysicochemical properties of water (Franks 1973, 1975a,b; Israelachvili 1992; Fennema1996b). It is the ability of water molecules to form relatively strong bonds with each otherand with other types of polar or ionic molecules which determines many of the characteristicproperties of food emulsions.


4.3.2. Bulk Physicochemical Properties

The bulk physicochemical properties of pure water are determined by the mass, dimensions,bond angles, charge distribution, and interactions of the water molecule (Eisenberg andKauzmann 1969, Duckworth 1975, Simatos and Multon 1985, Fito et al. 1994). Water hasa high dielectric constant because the uneven distribution of partial charges on the moleculemeans that it is easily polarized by an electric field (Hasted 1972). It has a high melting point,boiling point, enthalpy of vaporization and surface tension compared to other molecules ofa similar size that also contain hydrogen (e.g., CH4, NH3, HF, and H2S), because a greateramount of energy must be supplied to disrupt the strong hydrogen bonds holding the watermolecules together in the condensed state (Israelachvili 1992, Fennema 1996b). Ice andliquid water have a relatively low density because the water molecules adopt a structure inwhich they are in direct contact with only four of their nearest neighbors, rather than forminga more closely packed structure (Franks 1975b, Fennema 1996b). The relatively low viscosityof water is due to the highly dynamic nature of hydrogen bonds compared to the time scaleof a rheology experiment. Even though energy is required to break the hydrogen bondsbetween water molecules as they move past each other, most of this energy is regained whenthey form new hydrogen bonds with their new neighbors.

The crystallization of water has a pronounced effect on the bulk physiochemical propertiesof food emulsions. The presence of ice crystals in the aqueous phase of an oil-in-wateremulsion, such as ice cream, contributes to the characteristic mouthfeel and texture of theproduct (Berger 1976). When these ice crystals grow too large, a product is perceived asbeing “grainy” or “sandy,” which is commonly experienced when ice cream is melted andthen refrozen (Berger 1976). Many foods are designed to be freeze–thaw stable (i.e., thequality should not be adversely affected once the product is frozen and then thawed) (Partmann1975). Considerable care must be taken in the choice of ingredients and freezing/thawingconditions to create a food emulsion which is freeze–thaw stable. The basic principles of icecrystallization are similar to those described for fats and oils (Section 4.2.3.). Nevertheless,water does exhibit some anomalous behavior because of its unique molecular properties (e.g.,it expands when it crystallizes, whereas most other substances contract) (Fennema 1996b).This is because the increased mobility of the water molecules in the liquid state means thatthey can get closer together, and so the density of the liquid state is actually greater than thatof the solid state. Some of the most important bulk physicochemical properties of liquid waterare compared with those of a liquid oil in Table 4.3. A more detailed discussion of themolecular basis of the physicochemical properties of water in relation to food quality is givenby Fennema (1996b).

4.4. AQUEOUS SOLUTIONS

The aqueous phase of most food emulsions contains a variety of water-soluble constituents,including minerals, acids, bases, flavors, preservatives, vitamins, sugars, surfactants, pro-teins, and polysaccharides (Dickinson 1992). Interactions between these constituents andwater molecules determine the solubility, partitioning, volatility, conformation, and chemicalreactivity of many food ingredients. It is therefore important for food scientists to understandthe nature of solute–water interactions and their influence on the bulk physicochemical andorganoleptic properties of food emulsions.

When a solute molecule is introduced into pure water, the normal structural organizationand interactions of the water molecules are altered, which results in changes in waterproperties such as density, compressibility, melting point, and boiling point (Franks 1973,Reichardt 1988, Israelachvili 1992, Murrell and Jenkins 1994, Fennema 1996b). The extent


of these changes depends on the molecular characteristics of the solute (i.e., its size, shape,and polarity). The water molecules in the immediate vicinity of the solute experience thelargest modification of their properties and are often referred to as being “bound” to thesolute (Reichardt 1988, Murrell and Jenkins 1994). In reality, these water molecules are notpermanently bound to the solute but rapidly exchange with the bulk water molecules, albeitwith a reduced mobility (Franks 1991, Fennema 1996b). The mobility of “bound” waterincreases as the strength of the attractive interactions between it and the solute decreases(i.e., nonpolar–water > dipole–water > ion–water) (Israelachvili 1992). The amount of water“bound” to a solute can be defined as the number of water molecules whose properties aresignificantly altered by its presence. In practice, it is difficult to unambiguously define orstipulate the amount of “bound” water (Franks 1991), first because the water molecules“bound” to a solute do not all have the same properties (the water molecules closest to thesolute are more strongly influenced by its presence than those farthest away) and, second,because each of the physicochemical properties that are measured in order to determine theamount of “bound” water is influenced to a different extent (e.g., density, compressibility,mobility, melting point). As a consequence, different analytical techniques often measuredifferent amounts of “bound” water, depending on the physical principles on which theyoperate.

4.4.1. Interaction of Water with Ionic Solutes

Many of the solutes present in food emulsions are either ionic or capable of being ionized,including salts, acids, bases, proteins, and polysaccharides (Fennema 1996b). The degreeof ionization of many of these solutes is governed by the pH of the aqueous solution, andso their interactions are particularly sensitive to pH. The ion–dipole interactions whichoccur between an ionic solute and a water molecule are usually stronger than the dipole–dipole interactions which occur between a pair of water molecules (Table 4.4). As a conse-quence, the water molecules in the immediate vicinity of an ion tend to orientate themselvesso that their oppositely charged dipole faces the ion. Thus, a positively charged ion causesthe water molecules to align themselves so that a δ– group faces the ion, whereas theopposite is true for a negatively charged ion (Figure 4.10). The relatively strong nature ofion–dipole interactions means that the mobility of the water molecules near the surface ofan ion is significantly less than that of bulk water (Reichardt 1988, Fennema 1996b,Robinson et al. 1996). The residence time of a water molecule in the vicinity of an ionicgroup is ≈10–8 s, whereas it is ≈10–11 s in bulk water (Fennema 1996b). The influence of anion on the mobility and alignment of the water molecules is greatest at its surface becausethe electric field is strongest there. As one moves away from the ion surface, the strengthof the electric field decreases, so that the ion–water interactions become progressivelyweaker. Thus, the water molecules become more mobile and are less likely to be aligned

TABLE 4.4Typical Water–Solute Interactions Found in Food Emulsions

Interaction Strength of interactiontype Typical example compared to water–H bond

Water–ion Free ion (e.g., Na+, Cl–) GreaterIonic group (e.g., –CO2

–, –NH3+)

Water–dipole –C=O, –NH, –OH SimilarWater–nonpolar Alkyl group Much smaller


FIGURE 4.10 Organization of water molecules around ions in aqueous solutions.

toward the ion. At a sufficiently large distance from the ion surface, the water molecules areuninfluenced by its presence and have properties similar to those of bulk water (Franks1973, Reichardt 1988). Alterations in the structural organization and interactions of watermolecules in the vicinity of an ion cause significant changes in the physicochemical prop-erties of water (Fennema 1996b). The water that is “bound” to an ionic solute is less mobile,less compressible, more dense, has a lower freezing point, and has a higher boiling pointthan bulk water. Most ionic solutes have a high water solubility because the formation ofmany ion–dipole bonds in an aqueous solution helps to compensate for the loss of the strongion–ion bonds in the crystals, which is coupled with the favorable entropy of mixingcontribution (Chapter 2).

The number of water molecules whose mobility and structural organization are altered bythe presence of an ion increases as the strength of its electric field increases. The strength ofthe electrical field generated by an ion is determined by its charge divided by its radius(Israelachvili 1992). Thus, ions which are small and/or multivalent generate strong electricfields that influence the properties of the water molecules up to relatively large distances fromtheir surface (e.g., Li+, Na+, H3O+, Ca2+, Ba2+, Mg2+, Al3+ and OH–). On the other hand, ionswhich are large and/or monovalent generate relatively weak electrical fields, and thereforetheir influence extends a much shorter distance into the surrounding water (e.g., K+, Rb+, Cs+,NH4

+, Cl–, Br–, and I–). The number of water molecules “bound” to an ion is usually referredto as the hydration number. Thus the hydration number of small multivalent ions is usuallylarger than that of large monovalent ions.

When an ionic solute is added to pure water, it disrupts the existing tetrahedral arrangementof the water molecules but imposes a new order on the water molecules in its immediatevicinity (Franks 1973, Reichardt 1988). The overall structural organization of the watermolecules in an aqueous solution can therefore either increase or decrease after a solute isadded, depending on the amount of structure imposed on the water by the ion compared tothat lost by disruption of the tetrahedral structure of bulk water (Israelachvili 1992). If thestructure imposed by the ion is greater than that lost by the bulk water, the overall structuralorganization of the water molecules is increased, and the solute is referred to as a structuremaker (Robinson et al. 1996). Ionic solutes that generate strong electric fields are structuremakers, and the magnitude of their effect increases as the size of the ions decreases and/ortheir charge increases. If the structure imposed by an ion is not sufficiently large to compen-sate for that lost by disruption of the tetrahedral structure of bulk water, then the overallstructural organization of the water molecules in the solution is decreased, and the solute isreferred to as a structure breaker (Robinson et al. 1996). Ionic solutes that generate weakelectric fields are structure breakers, and the magnitude of their effect increases as their sizeincreases or they become less charged (Israelachvili 1992).


The influence of ionic solutes on the overall properties of water depends on their concen-tration (Israelachvili 1992, Robinson et al. 1996). At low solute concentrations, the majorityof water is not influenced by the presence of the ions and therefore has properties similar tothat of bulk water. At intermediate solute concentrations, some of the water molecules haveproperties similar to those of bulk water, whereas the rest have properties which are domi-nated by the presence of the ions. At high solute concentrations, all the water molecules maybe influenced by the presence of the solute molecules and therefore have properties which areappreciably different from those of bulk water. The solubility of biopolymer molecules inaqueous solutions decreases as the concentration of ionic solutes increases above a certainlevel, which is known as “salting out,” because the solutes compete with the biopolymers forthe limited amount of water that is available to hydrate them.

The presence of electrolytes in the aqueous phase of an emulsion can influence theinteractions between the droplets in a variety of ways. First, ions can screen electrostaticinteractions (Section 3.4.3) or the zero-frequency contribution of the van der Waals interac-tion (Section 3.3.4). Second, multivalent ions may be able to form salt bridges betweencharged droplets (Section 3.4.4). Third, the ability of ions to alter the structural organizationof water influences the strength of hydrophobic interactions (Section 3.7). Fourth, ions cancause changes in the size of biopolymer molecules in solution, which alters the strength ofsteric and depletion interactions (Sections 3.5 and 3.6). Finally, the binding of hydrated ionsto the surface of emulsion droplets increases the hydration repulsion between them (Section3.8). The fact that ions influence the interactions between emulsion droplets in many differentways means that it is difficult to accurately quantify their effect on emulsion properties.

4.4.2. Interaction of Water with Dipolar Solutes

Many food ingredients contain molecules which are either dipolar or contain dipolar groups,including alcohols, sugars, proteins, polysaccharides, and surfactants (Fennema 1996a). Wateris capable of participating in dipole–dipole interactions with dipolar solutes (Franks 1973,1991). By far the most important type of dipole–dipole interaction is between water and thosesolutes which have hydrogen bond donors (e.g., –O–Hδ+) or acceptors (e.g., δ–O–). Thestrength of hydrogen bonds between water and this type of dipolar solute is similar to thatbetween two water molecules (Table 4.4). The addition of a dipolar solute to water thereforehas much less influence on the mobility and organization of the water molecules in itsimmediate vicinity than does a similarly sized ionic solute (Fennema 1996b). The influenceof dipolar solutes on the properties of water is largely governed by the ease with which theycan be accommodated into the existing tetrahedral structure of the water molecules (Figure4.11). When a dipolar solute is of an appropriate size and shape, and has hydrogen bondacceptors and donors at positions where they can easily form bonds with the neighboringwater molecules, it can fit into the tetrahedral structure (Brady and Ha 1975, Galema andHoiland 1991, Franks 1991, Schmidt et al. 1994). For this type of solute, there need be littlechange in the number of hydrogen bonds formed per water molecule or the overall structuralorganization of the water molecules. This type of solute therefore tends to be highly watersoluble because of the entropy of mixing (Chapter 2). If the solute molecule is not of anappropriate size and shape, or if its hydrogen bond donors and acceptors are not capable ofaligning with those of neighboring water molecules, then it cannot easily fit into the tetrahe-dral structure of water (Figure 4.11). This causes a dislocation of the normal water structuresurrounding the solute molecules, which is energetically unfavorable. In addition, there maybe a significant alteration in the physicochemical properties of the water molecules in thevicinity of the solute. For this reason, dipolar solutes which are less compatible with thetetrahedral structure of water tend to be less soluble than those which are compatible.


FIGURE 4.11 The effect of dipolar solutes on the structural organization of water molecules dependson their dimensions and the number and position of their hydrogen bonding groups.

Just as with ionic solutes, the effect of dipolar solutes depends on their concentration. Atlow solute concentrations, most of the water has the same properties as bulk water, but at highconcentrations, a significant proportion of the water has properties which are altered by thepresence of the solute. Nevertheless, it takes a greater concentration of a dipolar solute tocause the same effect as an ionic solute because of the greater strength of ion–water inter-actions compared to dipole–water interactions.

Interactions between dipolar groups and water determine a number of important propertiesof food components in emulsions. The hydration of the dipolar head groups of surfactantmolecules is believed to be partly responsible for their stability to aggregation (Evans andWennerstrom 1994). When surfactants are heated, the head groups become progressivelydehydrated, which eventually causes the molecules to aggregate (Section 4.5). These hydra-tion forces also play an important role in preventing the aggregation of emulsion dropletsstabilized by nonionic surfactants (Section 3.8). The three-dimensional conformation andinteractions of proteins and polysaccharides are influenced by their ability to form intramo-lecular and intermolecular hydrogen bonds (Section 4.6). The solubility, partitioning, andvolatility of dipolar solutes depend on their molecular compatibility with the surroundingsolvent: the stronger the molecular interactions between a solute and its neighbors in a liquid,the greater its solubility and the lower its volatility (Baker 1987).

4.4.3. Interaction of Water with Nonpolar Solutes:The Hydrophobic Effect

The attraction between a water molecule and a nonpolar solute is much weaker than thatbetween two water molecules, because nonpolar molecules are incapable of forming hydro-gen bonds (Israelachvili 1992, Evans and Wennerstrom 1994). For this reason, when a


FIGURE 4.12 Schematic representation of the reorganization of water molecules near a nonpolarsolute.

nonpolar molecule is introduced into pure liquid water, the water molecules surrounding itchange their orientation so that they can maximize the number of hydrogen bonds formedwith neighboring water molecules (Figure 4.12). The structural rearrangement and alterationin the physicochemical properties of water molecules in the immediate vicinity of a nonpolarsolute is known as hydrophobic hydration. At relatively low temperatures, it is believed thata “cage-like” or clathrate structure of water molecules exists around a nonpolar solute, inwhich the water molecules involved have a coordination number of four, which is greaterthan that of the water molecules in the bulk phase (3 to 3.5) (Israelachvili 1992). Despitegaining some order, the water molecules in the cage-like structures are still highly dynamic,having residence times of the order of 10–11 s (Evans and Wennerstrom 1994). The alterationin the organization and interactions of water molecules surrounding a nonpolar solute hasimportant implications for the solubility and interactions of nonpolar groups in water (Tanford1980, Dill 1990, Israelachvili 1992, Cramer and Tuhlar 1994, Fennema 1996b).

The behavior of nonpolar solutes in water can be understood by considering the transferof a nonpolar molecule from an environment where it is surrounded by similar molecules toone where it is surrounded by water molecules (Tanford 1980). When a nonpolar solute istransferred from a nonpolar solvent into water, there are changes in both the enthalpy(∆Htransfer) and entropy (∆Stransfer) of the system. The enthalpy change is related to the alter-ation in the overall strength of the molecular interactions, whereas the entropy change isrelated to the alteration in the structural organization of the solute and solvent molecules. Theoverall free energy change (∆Gtransfer) depends on the relative magnitude of these two contri-butions (Evans and Wennerstrom 1994):

∆ ∆ ∆G H T Stransfer transfer transfer= − (4.7)

The relative contribution of the enthalpic and entropic contributions to the free energydepends on temperature (Table 4.5). An understanding of the temperature dependence of thefree energy of transfer is important for food scientists because it governs the behavior ofmany food components during food processing, storage, and handling. At relatively lowtemperatures (<15°C), the number of hydrogen bonds formed by the water molecules in thecage-like structure surrounding the nonpolar solute is slightly higher than in bulk water, andso ∆Htransfer is negative (i.e., favors transfer). On the other hand, the water molecules in directcontact with the nonpolar solute are more ordered than those in bulk water, and so the entropyterm is positive (i.e., opposes transfer). Overall, the entropy term dominates, and so thetransfer of a nonpolar molecule into water is thermodynamically unfavorable (Tanford 1980,Israelachvili 1992).


TABLE 4.5Temperature Dependence of the Enthalpic and Entropic Contributionsto the Free Energy Change That Occurs When a Nonpolar Molecule(Benzene) Is Transferred from the Neat Liquid to Water

T (C) H (kJ mol–1) –TS (J C–1 mol–1) G (kJ mol–1)

15 –0.15 18.89 18.7525 2.08 17.30 19.3830 3.16 16.50 19.6635 4.37 15.55 19.92

Adapted from Baldwin 1986.

As the temperature is raised, the water molecules become more thermally agitated, and sotheir organization within the cage-like structure is progressively lost, which has consequencesfor both the enthalpy and entropy contributions. First, some of the partial charges on the watermolecules face toward the nonpolar group and are therefore unable to form hydrogen bondswith the surrounding water molecules. Thus, the number of hydrogen bonds formed by thewater molecules in the cage-like structure decreases with increasing temperature. At a certaintemperature, the number of hydrogen bonds formed by the water molecules in the cage-likestructure becomes less than that of bulk water. Below this temperature, the enthalpy associ-ated with transferring a nonpolar molecule into water is negative (favorable to transfer), butabove it, it is positive (unfavorable to transfer). The enthalpy term therefore makes anincreasing contribution to opposing the transfer of nonpolar molecules into water as thetemperature rises. Second, the increasing disorganization of the water molecules surroundinga nonpolar molecule as the temperature is raised means that the entropy difference betweenthe water molecules in the cage-like structure and those in the bulk water is lessened. Thus,as the temperature is increased, the contribution of the entropy term becomes progressivelyless important. In summary, at low temperatures, the major contribution to the unfavorablefree energy change associated with transfer of a nonpolar molecule into water is the entropyterm, but at higher temperatures it is the enthalpy term. Overall, the transfer of a nonpolarmolecule from an organic solvent into water becomes increasingly thermodynamically unfa-vorable as the temperature is raised (Table 4.5).

The free energy associated with transferring a nonpolar molecule from an environmentwhere it is surrounded by similar molecules to one in which it is surrounded by watermolecules has been shown to be a product of its surface area and the interfacial tensionbetween the bulk nonpolar liquid and water (i.e., ∆G = γ∆A) (Tunon et al. 1992). An aqueoussolution containing a nonpolar solute can decrease its free energy by reducing the unfavorablecontact area between the nonpolar groups and water, which is known as the hydrophobiceffect. The strong tendency for nonpolar molecules to associate with each other in aqueoussolutions is a result of the attempt of the system to reduce the contact area between water andnonpolar regions and is known as the hydrophobic interaction (Section 3.7). The hydrophobiceffect is responsible for many of the characteristic properties of food emulsions, including theaggregation of proteins, the formation of surfactant micelles, the adsorption of emulsifiers atoil–water and air–water interfaces, the aggregation of hydrophobic particles, and the immis-cibility of oil and water.

The strength of the hydrophobic interaction between nonpolar substances in water isaffected by the presence of ions in the aqueous phase separating them. Ions can either increaseor decrease the structural organization of water molecules in an aqueous solution, depending


on whether they are structure makers or structure breakers (Section 4.4.1). As one of themajor driving forces for hydrophobic interactions is the difference in structural organization(entropy) between the water molecules in the immediate vicinity of the nonpolar solute andthose in the bulk water, then changes in the organization of the water molecules in bulk wateralter its strength (Israelachvili 1992). Structure makers decrease the magnitude of the hydro-phobic interaction and therefore increase the water solubility of nonpolar solutes because thedifference in structural organization is reduced, whereas structure breakers have the oppositeeffect. The strength of the hydrophobic interaction also depends on temperature, increasingas the temperature is raised.

4.5. SURFACTANTS

4.5.1. Molecular Characteristics

The principal role of surfactants in food emulsions is to enhance their formation and stability(Charalambous and Doxastakis 1989, Dickinson 1992, Hasenhuettl and Hartel 1997); how-ever, they may also alter emulsion properties in a variety of other ways (e.g., by interactingwith proteins or polysaccharides, by forming surfactant micelles, or by modifying thestructure of fat crystals (Dickinson and McClements 1995, Bergenstahl 1997, Bos et al.1997, Deffenbaugh 1997). By definition, a surfactant is an amphiphilic molecule that has ahydrophilic “head” group which has a high affinity for water and a lipophilic “tail” groupwhich has a high affinity for oil (Myers 1988, Hasenhuettl 1997). Surfactants can thereforebe represented by the formula RX, where X represents the hydrophilic head and R thelipophilic tail (Dickinson and McClements 1995). The characteristics of a particular surfac-tant depend on the nature of its head and tail groups (Table 4.6). The head group may beanionic, cationic, zwitterionic, or nonionic (Myers 1988, St. Angelo 1989, Zielinski 1997).Surfactants used in the food industry are mainly nonionic (e.g., monoacylglycerols, sucroseesters of fatty acids), anionic (e.g., fatty acids), or zwitterionic* (e.g., lecithin). The tailgroup usually consists of one or more hydrocarbon chains, with between 10 and 20 carbonatoms per chain (St. Angelo 1989, Bergenstahl 1997, Zielinski 1997). The chains may besaturated or unsaturated, linear or branched, aliphatic and/or aromatic. Most surfactants usedin foods have either one or two linear aliphatic chains, which may be saturated or unsatur-ated. Each type of surfactant has functional properties that are determined by its uniquechemical structure (Myers 1988). Thus, there is no single surfactant which is appropriate forevery application. Instead, it is necessary to choose a surfactant which is the most appropri-ate for each particular application.

Surfactants aggregate spontaneously in solution to form a variety of thermodynamicallystable structures known as association colloids (e.g., micelles, bilayers, vesicles, and reversemicelles) (Figure 4.13). These structures are adopted because they minimize the unfavorablecontact area between the nonpolar tails of the surfactant molecules and water (Hiemenz 1986,Evans and Wennerstrom 1994). The type of association colloid formed by a surfactantdepends principally on its polarity and molecular geometry (Section 4.5.3). Associationcolloids are held together by physical interactions that are relatively weak compared to thethermal energy, and therefore they have highly dynamic and flexible structures (Israelachvili1992). In addition, their structures are particularly sensitive to changes in environmentalconditions, such as temperature, pH, ionic strength, and ion type (MacKay 1987, Myers 1988,Israelachvili 1992). Because surfactant micelles are the most important type of association

* A zwitterionic surfactant is one which has both positively and negatively charged groups on the same molecule.


TABLE 4.6Selected HLB Group Numbers

Hydrophilic group Group number Lipophilic group Group number

–SO4Na+ 38.7 –CH– 0.475–COO–H+ 21.2 –CH2– 0.475Tertiary amine 9.4 –CH3 0.475Sorbitan ring 6.8–COOH 2.1–O– 1.3

Adapted from Davis 1994b.

FIGURE 4.13 Some typical structures formed due to the self-association of surfactant molecules.

colloid formed in many food emulsions, we will focus principally on their properties (Dickinsonand McClements 1995).

4.5.2. Functional Properties

4.5.2.1. Critical Micelle Concentration

A surfactant forms micelles in an aqueous solution when its concentration exceeds somecritical level, known as the critical micelle concentration or CMC (Myers 1988). Below theCMC, the surfactant molecules are dispersed predominantly as monomers, but once the CMCis exceeded, any additional surfactant molecules form micelles, and the monomer concentra-tion remains fairly constant (Hiemenz 1986). Despite the highly dynamic nature of theirstructure, surfactant micelles have a fairly well-defined average size and shape under a givenset of environmental conditions. Thus, when surfactant is added to a solution above the CMC,the number of micelles tends to increase, rather than the size or shape of the individualmicelles (although this may not be true at high surfactant concentrations). There is an abruptchange in the physicochemical properties of a surfactant solution when the CMC is exceeded(e.g., surface tension, electrical conductivity, turbidity, and osmotic pressure) (Rosen 1978,Hiemenz 1986). This is because the properties of surfactant molecules dispersed as mono-mers are different from those in micelles. For example, surfactant monomers are amphiphilic


and have a high surface activity, whereas micelles have little surface activity because theirsurface is covered with hydrophilic head groups. Consequently, the surface tension of asolution decreases with increasing surfactant concentration below the CMC, but remainsfairly constant above it (Chapter 5).

4.5.2.2. Cloud Point

When a surfactant solution is heated above a certain temperature, known as the cloud point,it becomes turbid (Myers 1988). This occurs because the hydrophilic head groups of thesurfactant molecules become progressively dehydrated as the temperature is raised, whichalters their molecular geometry (Section 4.5.3.3) and decreases the hydration repulsionbetween them (Israelachvili 1992, Evans and Wennerstrom 1994). Above a certain tempera-ture, known as the cloud point, the micelles form aggregates which are large enough to scatterlight and therefore make the solution appear turbid. As the temperature is increased further,the aggregates may grow so large that they sediment under the influence of gravity and forma separate phase which can be observed visually. The cloud point increases as the hydropho-bicity of a surfactant molecule increases (i.e., the length of its hydrocarbon tail increases orthe size of its hydrophilic head group decreases) (Myers 1988).

4.5.2.3. Solubilization

Nonpolar molecules, which are normally insoluble or only sparingly soluble in water, can besolubilized in an aqueous surfactant solution by incorporation into micelles or other types ofassociation colloids (Elworthy et al. 1968, Mukerjee 1979, MacKay 1987, Christian andScamehorn 1995). The resulting system is thermodynamically stable; however, it may takean appreciable time to reach equilibrium because of the time taken for molecules to diffusethrough the system and because of the activation energy associated with transferring anonpolar molecule from a bulk phase into a micelle (Dickinson and McClements 1995,Kabalnov and Weers 1996). Micelles containing solubilized materials are referred to asswollen micelles or microemulsions, whereas the material solubilized within the micelle isreferred to as the solubilizate. The ability of micellar solutions to solubilize nonpolar mol-ecules has a number of potentially important applications in the food industry, includingselective extraction of nonpolar molecules from oils, controlled ingredient release, incorpo-ration of nonpolar substances into aqueous solutions, transport of nonpolar molecules acrossaqueous membranes, and modification of chemical reactions (Dickinson and McClements1995). There are three important factors which determine the functional properties of swollenmicellar solutions: (1) the location of the solubilizate within the micelles, (2) the maximumamount of material that can be solubilized per unit mass of surfactant, and (3) the rate atwhich solubilization proceeds. Food manufacturers must therefore select a micellar systemwhich is most appropriate for their particular application.

4.5.2.4. Surface Activity and Droplet Stabilization

Surfactants are used widely in the food industry to enhance the formation and stability of foodemulsions (St. Angelo 1989, Dickinson 1992, Bergenstahl 1997). To do this, they mustadsorb to the surface of emulsion droplets during homogenization and form a protectivemembrane which prevents the droplets from aggregating with each other during a collision(Walstra 1993a, 1996a,b). Surfactant molecules adsorb to oil–water interfaces because theycan adopt an orientation in which the hydrophilic part of the molecule is located in the waterwhile the hydrophobic part is located in the oil. This minimizes the contact area betweenhydrophilic and hydrophobic regions and therefore reduces the interfacial tension (Chapter


5). This reduction in interfacial tension is important during homogenization because it facili-tates the further disruption of emulsion droplets (i.e., less energy is needed to break up adroplet when the interfacial tension is lowered) (Chapter 6). Once adsorbed to the surface ofa droplet, the surfactant must provide a repulsive force which is strong enough to prevent thedroplet from aggregating with its neighbors (Chapters 3 and 7). Ionic surfactants providestability by causing all the emulsion droplets to have the same electric charge and thereforeelectrostatically repel each other. Nonionic surfactants provide stability by generating anumber of short-range repulsive forces which prevent the droplets from getting too closetogether, such as steric, hydration, and thermal fluctuation interactions. Some surfactantsform multilayers (rather than monolayers) at the surface of an emulsion droplet, which greatlyenhances the stability of the droplets against aggregation (Friberg and El-Nokaly 1983). Insummary, surfactants must have three characteristics to be effective at enhancing the forma-tion and stability of emulsions (Chapter 6). First, they must rapidly adsorb to the surface ofthe freshly formed emulsion droplets during homogenization. Second, they must reduce theinterfacial tension by a significant amount. Third, they must form an interfacial layer thatprevents the droplets from aggregating. It should also be noted that the ability of surfactantsto form micelles in the continuous phase of an emulsion can have a negative impact onemulsion stability, because they induce depletion flocculation or facilitate the transport of oilmolecules between droplets (Dickinson and McClements 1995). The ability of surfactants toregulate the interactions between droplets can also have a pronounced influence on therheological properties of emulsions (Chapter 8).

4.5.3. Surfactant Classification

A food manufacturer must consider a variety of factors when selecting a surfactant for aparticular product, including its legal status as a food ingredient; its cost; the reliability of thesupplier; the consistency of its quality from batch to batch; its ease of handling and disper-sion; its shelf life; its compatibility with other ingredients; the processing, storage, andhandling conditions it will experience; and the expected shelf life and physicochemicalproperties of the final product. How does a food manufacturer decide which surfactant is mostsuitable for a product? There have been various attempts to develop systems to classifysurfactants according to their physicochemical properties. Classification schemes have beendeveloped which are based on a surfactant’s solubility in oil and/or water (Bancroft’s rule),its ratio of hydrophilic to lipophilic groups (HLB number), and its molecular geometry (Davis1994b, Dickinson and McClements 1995, Bergenstahl 1997). Ultimately, all of these prop-erties depend on the chemical structure of the surfactant, and so the different classificationschemes are closely related.

4.5.3.1. Bancroft’s Rule

One of the first empirical rules developed to describe the type of emulsion that could bestabilized by a given surfactant was proposed by Bancroft (Davis 1996). Bancroft’s rule statesthat the phase in which the surfactant is most soluble will form the continuous phase of anemulsion. Hence a water-soluble surfactant should stabilize oil-in-water emulsions, whereasan oil-soluble surfactant should stabilize water-in-oil emulsions. This rule is applicable to awide range of surfactants, although there are many important exceptions. For example, manyamphiphilic molecules are highly soluble in either one phase or the other, but do not formstable emulsions because they are not particularly surface active or do not protect dropletsagainst aggregation. Bancroft’s rule is a useful empirical method for classifying surfactants;however, it provides little insight into the relationship between the molecular structure of asurfactant and its ability to form or stabilize emulsions.


4.5.3.2. Hydrophile–Lipophile Balance

The hydrophile–lipophile balance (HLB) concept is a semiempirical method which is widelyused for classifying surfactants. The HLB is described by a number which gives an indicationof the relative affinity of a surfactant molecule for the oil and aqueous phases (Davis 1994b).Each surfactant is assigned an HLB number according to its chemical structure. A moleculewith a high HLB number has a high ratio of hydrophilic groups to lipophilic groups and viceversa. The HLB number of a surfactant can be calculated from a knowledge of the numberand type of hydrophilic and lipophilic groups it contains, or it can be estimated fromexperimental measurements of its cloud point (Shinoda and Friberg 1986). The HLB numbersof many surfactants have been tabulated in the literature (Shinoda and Kunieda 1983; Becher1985, 1996). A widely used semiempirical method for calculating the HLB number of asurfactant is as follows (Davis 1994b):

HLB hydrophilic group numbers) ± lipophilic group numbers)= +7 Σ Σ( ( (4.8)

Group numbers have been assigned to many different types of hydrophilic and lipophilicgroups (Table 4.6). The sums of the group numbers of all the lipophilic groups and of allthe hydrophilic groups are substituted into Equation 4.8 and the HLB number is calculated.Despite originally being developed as a semiempirical equation, Equation 4.8 has beenshown to have a thermodynamic basis, with the sums corresponding to the free energychanges in the hydrophilic and lipophilic parts of the molecule when micelles are formed(Becher 1985).

The HLB number of a surfactant gives a useful indication of its solubility in either the oiland/or water phase and can be used to predict the type of emulsion that will be formed(Davis 1994b). A surfactant with a low HLB number (3 to 6) is predominantly hydrophobic,dissolves preferentially in oil, stabilizes water-in-oil emulsions, and forms reverse micellesin oil. A surfactant with a high HLB number (8 to 18) is predominantly hydrophilic,dissolves preferentially in water, stabilizes oil-in-water emulsions, and forms micelles inwater. A surfactant with an intermediate HLB number (6 to 8) has no particular preferencefor either oil or water. Molecules with HLB numbers below 3 and above 18 are notparticularly surface active and tend to accumulate preferentially in bulk oil or aqueousphases, rather than at an oil–water interface. Emulsion droplets are particularly prone tocoalescence when they are stabilized by surfactants that have extreme or intermediate HLBnumbers. At very high or low HLB numbers, a surfactant has such a low surface activity thatit does not accumulate appreciably at the droplet surface and therefore does not provideprotection against coalescence. At intermediate HLB numbers (6 to 8), emulsions areunstable to coalescence because the interfacial tension is so low that very little energy isrequired to disrupt the membrane. Maximum stability of emulsions is obtained for oil-in-water emulsions using surfactants with an HLB number around 10 to 12 and for water-in-oil emulsions around 3 to 5. This is because the surfactants are surface active but do notlower the interfacial tension so much that the droplets are easily disrupted. It is possible toadjust the effective HLB number by using a combination of two or more surfactants withdifferent HLB numbers (Becher 1957).

One of the major drawbacks of the HLB concept is that it does not take into account thefact that the functional properties of a surfactant molecule are altered significantly bychanges in temperature or solution conditions (Davis 1994b). Thus a surfactant may becapable of stabilizing oil-in-water emulsions at one temperature but water-in-oil emulsionsat another temperature, even though it has exactly the same chemical structure. The HLBconcept could be extended to include temperature effects by determining the group numbers


as a function of temperature, although this would be a rather tedious and time-consumingtask.

4.5.3.3. Molecular Geometry and the Phase Inversion Temperature

The molecular geometry of a surfactant molecule can be described by a packing parameter,(p) (Israelachvili 1992, 1994; Kabalnov and Wennerstrom 1996):

pv

la=

0(4.9)

where v and l are the volume and length of the hydrophobic tail and a0 is the cross-sectionalarea of the hydrophilic head group (Figure 4.14). When surfactant molecules associate witheach other, they tend to form monolayers that have a curvature which allows the mostefficient packing of the molecules. At this optimum curvature, the monolayer has its lowestfree energy, and any deviation from this curvature requires the expenditure of energy. Theoptimum curvature (H0) of a monolayer depends on the packing parameter of the surfactant:for p = 1, monolayers with zero curvature (H0 = 0) are preferred; for p < 1, the optimumcurvature is convex (H0 < 0); and for p > 1, the optimum curvature is concave (H0 > 0) (Figure4.14). Simple geometrical considerations indicate that spherical micelles are formed when pis less than one-third, nonspherical micelles when p is between one-third and one-half, andbilayers when p is between one-half and one (Israelachvili 1992, 1994). Above a certainconcentration, bilayers join up to form vesicles because energetically unfavorable end effectscan be eliminated. At values of p greater than 1, reverse micelles are formed, in which thehydrophilic head groups are located in the interior (away from the oil) and the hydrophobictail groups are located at the exterior (in contact with the oil) (Figure 4.14). The packingparameter therefore gives a useful indication of the type of association colloid that a surfac-tant molecule forms in solution.

The packing parameter is also useful because it accounts for the temperature dependenceof the physicochemical properties of surfactant solutions and of emulsions (Kabalnov andWennerstrom 1996). The temperature at which a surfactant solution converts from a micellarto a reverse-micellar system or an oil-in-water emulsion changes to a water-in-oil emulsion

FIGURE 4.14 The physicochemical properties of surfactants can be related to their moleculargeometry.


is known as the phase inversion temperature or PIT (Shinoda and Kunieda 1983, Shinoda andFriberg 1986). Consider what happens when an emulsion that is stabilized by a surfactant isheated (Figure 4.15). At temperatures well below the PIT (≈20°C), the packing parameter issignificantly less than unity, and so a system that consists of an oil-in-water emulsion inequilibrium with a swollen micellar solution is favored. As the temperature is raised, thehydrophilic head groups of the surfactant molecules become progressively dehydrated, whichcauses p to increase toward unity. Thus the emulsion droplets become more prone to coales-cence and the swollen micelles grow in size. At the PIT, p = 1, and the emulsion breaks downbecause the droplets have an ultralow interfacial tension and therefore readily coalesce witheach other (Aveyard et al. 1990, Kabalnov and Weers 1996). The resulting system consistsof excess oil and excess water (containing some surfactant monomers), separated by a thirdphase that contains surfactant molecules aggregated into bilayer structures. At temperaturessufficiently greater than the PIT (≈20°C), the packing parameter is much larger than unity,and the formation of a system which consists of a water-in-oil emulsion in equilibrium withswollen reverse micelles is favored. A further increase in temperature leads to a decrease inthe size of the reverse micelles and in the amount of water solubilized within them. Themethod of categorizing surfactant molecules according to their molecular geometry is nowwidely accepted as the most useful means of determining the type of emulsions they tend tostabilize (Kabalnov and Wennerstrom 1996).

4.5.3.4. Other Factors

The classification schemes mentioned above provide information about the type of emulsionthat a surfactant tends to stabilize (i.e., oil-in-water or water-in-oil), but they do not providemuch insight into the size of the droplets which are formed during homogenization, theamount of surfactant required to form a stable emulsion, or the stability of the emulsiondroplets once formed. In choosing a surfactant for a particular application, these factors mustalso be considered. The speed at which a surfactant adsorbs to the surface of the emulsiondroplets produced during homogenization determines the minimum droplet size that can beproduced: the faster the adsorption rate, the smaller the size (Chapters 5 and 6). The amount

FIGURE 4.15 The phase inversion temperature occurs when the optimum curvature of a surfactantmonolayer is zero.


of surfactant required to stabilize an emulsion depends on the total surface area of thedroplets, as well as the surface area covered per unit mass of surfactant (Chapter 6). Themagnitude and range of the repulsive interactions generated by an interfacial surfactantlayer, as well as its viscoelasticity, determine the stability of emulsion droplets to aggrega-tion (Chapters 3 and 7).

4.6. BIOPOLYMERS

Proteins and polysaccharides are the two most important types of biopolymer used as ingre-dients in food emulsions (BeMiller and Whistler 1996, Damodaran 1996). Biopolymersprovide an important source of energy and essential nutrients in the human diet (Smolin andGrosvenor 1994). In addition, they have the ability to modify the appearance, texture,stability, and taste of food emulsions because of their unique functional characteristics (i.e.,their ability to stabilize emulsions and foams, to form gels, and to greatly enhance theviscosity of solutions) (Glicksman 1982a,b,c; Kinsella 1982; Dalgleish 1989; Damodaran1994, 1996; Dickinson 1992; Dickinson and McClements 1995). A large number of ingre-dient companies manufacture and supply biopolymer ingredients to the food industry. A widevariety of different types of biopolymer ingredient are available, each with its own uniquefunctional attributes. Selection of the most appropriate ingredient for a particular applicationis one of the most important decisions which a food manufacturer must make when creatinga food emulsion.

4.6.1. Molecular Characteristics

Proteins are polymers of amino acids, whereas polysaccharides are polymers of monosaccha-rides (MacGregor and Greenwood 1980, Lehninger et al. 1993, Damodaran 1996, BeMillerand Whistler 1996). The functional properties of biopolymers in foods are ultimately deter-mined by molecular characteristics, such as their molecular weight, conformation, flexibility,polarity, and interactions (Dea 1982, Kinsella 1982, Damodaran 1996, BeMiller and Whitler1996). These molecular characteristics are determined by the type, number, and sequence ofthe monomers that make up the polymer chain (van Holde 1977, MacGregor and Greenwood1980). Monomers vary according to their polarity (ionic, dipolar, nonpolar, or amphiphilic),dimensions, interactions, and functional groups (Lehninger et al. 1993). If a biopolymercontains only one type of monomer, it is referred to as a homopolymer (e.g., starch orcellulose), but if it contains different types of monomer, it is referred to as a heteropolymer(e.g., xanthan, gum arabic, and any naturally occurring protein).

Both proteins and polysaccharides have covalent linkages between the monomers aroundwhich the polymer chain can rotate (Lehninger et al. 1993). Because of this and the fact thatbiopolymers contain large numbers of monomers (typically between 20 and 20,000), they canpotentially take up a huge number of different configurations in solution. In practice, abiopolymer tends to adopt a conformation that minimizes its free energy under the prevailing

FIGURE 4.16 Biopolymers can adopt a number of different conformations in solution depending ontheir molecular structure. These can be conveniently categorized as random coil, rod-like, and globular.


environmental conditions.* A biopolymer does this by maximizing the number of favorableintermolecular and intramolecular interactions, minimizing the number of unfavorable ones,and maximizing its entropy, within the constraints set by its molecular structure and theprevailing environmental conditions (van Holde 1977, Dill 1990). Nevertheless, most foodsare nonequilibrium systems, and so a biopolymer may exist in a metastable state, becausethere is a large activation energy preventing it from reaching the most stable state. Theconfigurations that isolated biopolymers tend to adopt in aqueous solutions can be conve-niently divided into three categories: globular, rod-like, or random coil (Figure 4.16). Globu-lar biopolymers have fairly rigid compact structures, rod-like biopolymers have fairly rigidextended structures (often helical), and random-coil biopolymers have highly dynamic andflexible structures. Biopolymers can also be classified according to the degree of branchingof the chain (Lehninger et al. 1993). Most proteins have linear chains, whereas polysaccha-rides can have either linear (e.g., amylose) or branched (e.g., amylopectin) chains. In practice,many biopolymers do not have exclusively one type of conformation, but have some regionswhich are random coil, some which are rod-like, and some which are globular. In addition,biopolymer molecules may change from one type of conformation to another if their envi-ronment is altered (e.g., pH, ionic strength, solvent composition, or temperature): helix ⇔random coil or globular ⇔ random coil (Whitaker 1977, Sand 1982, Cesero 1994). Insolution, a biopolymer may exist as an isolated molecule, or it may be associated with othertypes of molecules. The molecular conformation and association of biopolymers are governedby a delicate balance of interaction energies and entropy effects (Dickinson and McClements1995).

4.6.2. Molecular Basis of Conformation and Aggregation

In this section, the most important types of molecular interaction and entropy effects whichinfluence the conformation and aggregation of biopolymer molecules in aqueous solution arediscussed.

4.6.2.1. Hydrophobic Interactions

Hydrophobic interactions play a major role in determining the molecular structure andaggregation of biopolymers which contain appreciable numbers of nonpolar groups (Alber1989, Dill 1990). Protein molecules often have significant amounts of nonpolar amino acidsalong their polypeptide backbone, such as valine, leucine, isoleucine, and phenylalanine(Lehninger et al. 1993). Most polysaccharides are predominantly hydrophilic, but some dohave nonpolar groups attached to their backbone (e.g., gum arabic and modified starch)(BeMiller and Whistler 1996). Biopolymers that have significant proportions of nonpolargroups tend to adopt an arrangement that minimizes the contact area between the nonpolargroups and water because of the hydrophobic effect (Section 4.4.3). This causes proteins thathave a large proportion of nonpolar amino acids to fold into compact globular structures inwhich the nonpolar amino acids are located in the interior of the protein (away from thewater), whereas the polar residues are located at the exterior (in contact with the water)(Damodaran 1996). Hydrophobic interactions are also believed to play an important role instabilizing the rod-like helical structures formed in some biopolymers, because nonpolargroups can be located in the core of the helix, away from the water.** Hydrophobic inter-

* Helix formation is also favored because it maximizes the number of hydrogen bonds formed.** It is possible that some biopolymers are “trapped” in a metastable state because there is a large energy barrier

between it and the lowest free energy state.


actions are also largely responsible for the aggregation and surfactivity of amphiphilicbiopolymers (Damodaran 1996). By aggregating to other nonpolar molecules, or by adsorbingto a nonpolar surface, a biopolymer can reduce the contact area between its hydrophobicgroups and water.

4.6.2.2. Electrostatic Interactions

Electrostatic interactions play a major role in determining the molecular structure and aggre-gation of many biopolymers (MacGregor and Greenwood 1980, Dickinson and McClements1995, Damodaran 1996). Proteins contain a number of amino acids that can ionize to formeither positively charged ions (e.g., arginine, lysine, proline, histidine, and the terminal aminogroup) or negatively charged ions (e.g., glutamic and aspartic acids and the terminal carboxylgroup) (Lehninger et al. 1993, Damodaran 1996). Some polysaccharides also have ionizablegroups on their backbone (e.g., sulfate, carboxylate, or phosphate groups) (MacGregor andGreenwood 1980, BeMiller and Whistler 1996). The net charge on a biopolymer moleculedepends on the pK values of its ionizable groups and the pH of the aqueous environment(Matthew 1985). Electrostatic interactions involving biopolymers are therefore particularlysensitive to the pH of the aqueous phase (Launay et al. 1986, Rha and Pradipasena 1986). Inaddition, they also depend on the concentration and type of counterions in the aqueous phasebecause of electrostatic screening effects (Section 3.4). If a biopolymer contains manysimilarly charged groups, it is more likely to adopt an extended configuration because thisincreases the average distance between the charges and therefore minimizes the unfavorableelectrostatic repulsions (Launay et al. 1986, Rha and Pradipasena 1986). If, on the other hand,the biopolymer contains many oppositely charged groups, it is more likely to fold up into acompact structure that maximizes the favorable electrostatic attractions. For this reason,proteins are often extremely compact at their isoelectric point and unfold as the pH is eitherincreased or decreased. Electrostatic interactions also play an important role in determiningthe aggregation of biopolymer molecules in solution. Similarly charged biopolymers repeleach other and therefore tend to exist as individual molecules, whereas oppositely chargedbiopolymers attract each other and therefore tend to aggregate (depending on the strength ofthe various other types of interactions involved). The binding of low-molecular-weight ions,such as Na+ and Ca2+, to biopolymer molecules is also governed by electrostatic interactionsand may influence the strength of the hydration repulsion between biopolymers in solution(Section 3.8).

4.6.2.3. Hydrogen Bonding

Both proteins and polysaccharides contain monomers that are capable of forming hydrogenbonds (Lehninger et al. 1993). Hydrogen bonds are a relatively strong type of molecularinteraction, and therefore a system attempts to maximize the number and strength of thehydrogen bonds formed. A biopolymer molecule may adopt an arrangement that enables itto maximize the number of hydrogen bonds which are formed between the monomers withinit, which leads to the formation of ordered regions such as helices, sheets, and turns (Lehningeret al. 1993). Alternatively, a biopolymer may adopt a less ordered structure where themonomers form hydrogen bonds with the surrounding water molecules. Thus, part or all ofa biopolymer may exist in either a highly ordered conformation (which is entropicallyunfavorable) with extensive intramolecular hydrogen bonding or may exist in a more ran-dom-coil conformation (which is entropically more favorable) with extensive intermolecularhydrogen bonding. The type of structure formed by a biopolymer under a certain set ofenvironmental conditions is governed by the relative magnitude of the hydrogen bonds


compared to the various other types of interactions, most notably hydrophobic, electrostatic,and configurational entropy.

Hydrogen bond formation is also responsible for the association of many types of foodbiopolymer. Segments of one biopolymer may be capable of forming strong hydrogen bondswith segments on another molecule, which causes the molecules to aggregate (Lehninger etal. 1993). These junction zones usually involve hydrogen-bonded helical or sheet-like struc-tures. Hydrogen-bonded junction zones tend to be stable at low temperatures but dissociateas the temperature is raised above a certain value because the configurational entropy termdominates (Section 4.6.2.7).

4.6.2.4. Steric Overlap and van der Waals Interactions

There is an extremely strong repulsive interaction between atoms or molecules at closeseparations because of the overlap of their electron clouds (Section 2.3.4). This stericoverlap interaction determines the minimum distance that two atoms or molecules can cometo each other. Steric overlap interactions greatly restrict the spatial distribution of themonomers within a polymer because two groups cannot occupy the same space. At themolecular level, van der Waals interactions operate between all types of molecules and tendto have fairly similar magnitudes (Israelachvili 1992). Consequently, they only play a minorrole in determining the conformation of biopolymers in solution, since there is little changein the van der Waals interactions in the folded and unfolded states. Nevertheless, if abiopolymer molecule is large enough to act as a colloidal particle, then there may be asignificantly strong van der Waals attraction, which favors its aggregation with other biopoly-mer molecules (Chapter 3).

4.6.2.5. Disulfide Bonds

The conformation and aggregation of many protein molecules are influenced by their abilityto form disulfide bonds (MacGregor and Greenwood 1980, Dickinson and McClements 1995,Damodaran 1996). Cystine is an amino acid that has a thiol group (–SH) which is capable offorming disulfide bonds (–S–S–) with other thiol groups by an oxidation reaction (Friedman1973). Free thiol groups may also participate in thiol–disulfide interchanges with disulfidebonds. Proteins are therefore capable of forming both intramolecular and intermoleculardisulfide bonds under the appropriate conditions. Intramolecular bonds form when a pair ofcysteine residues are brought into close proximity by the folding of a protein molecule andare important for stabilizing the structure of globular proteins against unfolding. Intermolecu-lar disulfide bonds form when two different protein molecules come close to each other andare important in determining the aggregation of proteins in gels and at interfaces (Dickinsonand Matsumura 1991, McClements et al. 1993d).

The chemistry of the sulfhydryl group (–SH) plays an important role in the molecularstructure and interactions of proteins (Friedman 1973). The pK value of the –SH group of freeL-cysteine is about 8.3. In proteins, the pK value is often higher and depends on the localenvironment of the –SH group and the degree of unfolding of the protein molecule (Friedman1973). The –S– ion, which predominates at high pH, is strongly nucleophilic and is capableof interacting with disulfide bonds by nucleophilic displacement.

P1–S–S–P2 + R–S– → P1–S– + R–S–S–P2 (4.10)

where P1 and P2 refer to polypeptide chains that may be on the same protein molecule or ondifferent protein molecules. The R–S– group may also be on the same or another protein


molecule, or it may even be some other chemical group (e.g., β-mercaptoethanol). If there isan excess of the reactive sulfhydryl group present, the disulfide bond between two polypep-tide chains may be broken:

P1–S–S–P2 + R–S–S–R → P1–S–S–R + R–S–S–P2 (4.11)

Disulfide interchange does not occur spontaneously in acidic conditions (pH <6) becausethe thiol group is below its pK value and is therefore not ionized (Friedman 1973). Thesulfyhydryl and disulfide groups in proteins are usually located in the interior of the foldedmolecule and are therefore unavailable for interaction, even under conditions that wouldnormally favor thiol/disulfide interchange. Unfolding of protein molecules exposes reactivesulfyhydryl groups and allows thiol–disulfide interchanges to take place under the appropri-ate conditions (Dickinson and Matsumura 1991, McClements et al. 1993d). This explains theincrease in sulfhydryl–disulfide interchange which occurs on heating proteins (McClementset al. 1994), when proteins are adsorbed to an interface (Dickinson and Matsumura 1991), orafter proteins are treated with proteolytic enzymes such as trypsin and chymotrypsin (Fried-man 1973).

4.6.2.6. Configurational Entropy

One of the most important factors determining the conformation and aggregation of biopoly-mer molecules in solution is their configurational entropy (Alber 1989, Dill 1990). Biopoly-mers have both local and nonlocal contributions to their configurational entropy. The localentropy refers to the number of conformations which sequences of monomers connected toone another along the chain can adopt. The formation of ordered structure, such as helicesor sheets, is entropically unfavorable. The nonlocal entropy is determined by the possibleconfigurations that the whole of the biopolymer chain can adopt. Thus, highly flexiblerandom-coil biopolymers, which can occupy a large number of different conformations, havea much higher configurational entropy than compact globular biopolymers, which can occupyfar fewer different conformations. Steric overlap interactions play an important role indetermining the magnitude of the nonlocal entropy. Two segments of a biopolymer chaincannot occupy the same volume, and therefore the number of configurations that the moleculecan adopt is restricted, which is entropically unfavorable.

The free energy associated with the configurational entropy (–T∆S) increases as the tem-perature is raised, which explains why globular proteins unfold at high temperatures, eventhough the hydrophobic attraction is stronger and therefore favors the folded state (Dickinsonand McClements 1995).

An important consequence of the nonlocal entropy contribution to the conformation of abiopolymer is the effect of covalent cross-links between different segments of the molecule.Introducing a covalent bond, such as a disulfide bond in a protein, severely restricts thenumber of conformations that the unfolded molecule can adopt and therefore reduces theentropic driving force that favors the random-coil conformation (Kinsella 1982).

4.6.2.7. Molecular Conformation and Aggregation

The conformation and aggregation of biopolymer molecules depend on the relative magni-tude of the various attractive and repulsive interactions which occur within and betweenmolecules, as well as their configurational entropy (Dickinson and McClements 1995).Consider a biopolymer that can exist in two different states:

State (1) State (2)↔ (4.12)


These states may be a folded and an unfolded state or an aggregated and an unaggregatedstate. The biopolymer will tend to exist in the state which has the lowest free energy underthe prevailing environmental conditions (e.g., pH, ionic strength, ion type, temperature, andsolvent type). The free energy of each state is governed by the various types of molecularinteractions and entropy contributions mentioned above. Thus, the overall free energy changewhich occurs when a biopolymer molecule undergoes a transition from one state to anothercan be represented by the following equation:

∆ ∆ ∆ ∆ ∆ ∆ ∆G G G G G G T Stransition VDW H HB S E CE= + + + + − (4.13)

where ∆GVDW, ∆GH, ∆GHB, ∆GS, and ∆GE are the differences in free energy between the twostates associated with van der Waals, hydrophobic, hydrogen bonding, steric, and electro-static interactions, and ∆SCE is the change in configurational entropy. If ∆Gtransition is nega-tive, the biopolymer will tend to undergo the transition; otherwise, it will not. It is conve-nient to group together the various contributions that favor the transition and those thatoppose it:

∆ ∆ ∆G G Gtransition oppose favor= + (4.14)

If the contributions that favor the transition dominate, then the biopolymer will move fromstate 1 to state 2; otherwise, it will remain in state 1 (Equation 4.12). For many systems, thefactors favoring the transition have a similar magnitude, but a different sign, than thoseopposing the transition (i.e., ∆Goppose ≈ –∆Gfavor) (Dickinson and McClements 1995). Thus themagnitude of ∆Gtransition is much smaller than either ∆Goppose or ∆Gfavor, because it is thedifference between these two relatively large numbers. This accounts for the fact that smallchanges in environmental conditions can cause a biopolymer to undergo a transition from onestate to another. For example, a globular protein may unfold when it is heated above a certaintemperature, or a random-coil protein may form a gel when it is cooled below a certaintemperature.

It must be stressed that most foods are nonequilibrium systems, and so the biopolymermolecules are not in their lowest free energy state. Often a biopolymer is “trapped” in ametastable state because there is a large activation energy which prevents it from reachingits equilibrium state (Section 1.2.1). For example, if a globular biopolymer is heated abovea temperature where it unfolds, it may aggregate with its neighbors. When the solution iscooled, the molecules are not able to disentangle themselves from their neighbors and so theyremain aggregated, even though a solution of individual molecules may have a lower freeenergy.

The conformation of a biopolymer in solution often determines its functional characteris-tics. For example, alterations in the conformation of β-lactoglobulin during the preparationof powdered whey-protein concentrates or isolates is one of the major reasons for theinconsistency of its quality from batch to batch. Thus, two ingredients that have exactly thesame composition may have very different functional properties because of alterations in thestructure of the proteins during processing.

4.6.3. Functional Properties

4.6.3.1. Protein Hydration and Water Solubility

Many of the functional properties of biopolymers in food emulsions are governed by theirinteractions with water (e.g., solubility, dispersibility, swelling, thickening, emulsification,


foaming, and gelling) (Suggett 1975a,b; Damodaran 1996; Fennema 1996b). Biopolymeringredients are usually added to the aqueous phase of food emulsions in a powdered form.The functional properties of many of these biopolymers are only exhibited when they arefully dissolved and evenly distributed throughout the aqueous phase. Consequently, theeffective dissolution of the powdered biopolymer ingredient in an aqueous solution is animportant part of emulsion preparation. This process involves a number of stages, includingdispersion, wetting, swelling, and dissolution. The effectiveness and rate of dissolutiondepend on many factors, including the pH, ionic strength, temperature, and composition ofthe aqueous phase, as well as the application of shearing forces.

It is useful to examine the way in which successive water molecules bind to dried biopoly-mer molecules (Fennema 1996b). First, the charged groups are hydrated, then the polargroups, and finally the nonpolar groups. Charged groups bind more water molecules pergroup (3 to 7) than polar groups (2 to 3), which in turn bind more than nonpolar groups (≈1).The water in a food containing biopolymers may exist in a number of different environments,including physically bound water, hydrodynamic water, capillary water, and free water(Fennema 1996b).

The water solubility of a biopolymer molecule depends on its compatibility with theaqueous solvent in which it is dispersed, which is governed by the relative magnitude ofbiopolymer–biopolymer, water–biopolymer, and water–water interactions. A biopolymer mol-ecule has a low water solubility when the strength of the water–biopolymer interactions issignificantly weaker than the average strength of the water–water and biopolymer–biopoly-mer interactions, because the molecules tend to associate with each other rather than with thesolvent molecules (Section 2.6). An appreciation of the factors which determine the watersolubility of a protein depends on an understanding of the various types of molecularinteraction and entropy effects discussed in Section 4.6.2.

In general, the water solubility of biopolymer molecules is determined by a combinationof van der Waals, electrostatic, hydrogen bonding, steric overlap, and hydrophobic interac-tions, as well as by the configurational entropy. In practice, the water solubility is usuallydominated by only one or two of these interactions, the most common being the hydrophobicand electrostatic interactions (Damodaran 1996). It is therefore important for food scientiststo identify the interactions which are most important for each type of biopolymer and toestablish the influence of solvent conditions on these interactions.

Hydrophobic interactions promote aggregation and therefore lead to poor water solubility.They are particularly important when biopolymer molecules have high proportions of non-polar groups on their surface. This accounts for the decreasing solubility of proteins withincreasing surface hydrophobicity (Damodaran 1996) and for the fact that many globularproteins become insoluble in water when they are heated above a temperature where theprotein molecule unfolds and exposes nonpolar groups (Damodaran 1996).

Electrostatic interactions play a major role in determining the water solubility of biopoly-mers that have charged groups and are important for all proteins and many polysaccharides(BeMiller and Whistler 1996, Damodaran 1996). Electrostatic interactions may be eitherattractive or repulsive, depending on the signs of the charges involved, and they may there-fore either increase or decrease water solubility. A protein molecule is positively charged atpH values below its isoelectric point and negatively charged at pH values above it. As aconsequence, there is an electrostatic repulsion between similarly charged molecules whichprevents them from coming close enough together to aggregate and therefore enhances theirwater solubility (de Wit and van Kessel 1996). In addition, the molecules are also preventedfrom aggregating because the ionized groups are highly hydrated, which leads to a short-range hydration repulsion (Section 3.8). The water solubility of most proteins decreasesdramatically at their isoelectric point (Franks 1991, Damodaran 1996). A protein molecule


has no net charge at its isoelectric point, but it does have some groups that are positivelycharged and some that are negatively charged. Proteins are therefore particularly susceptibleto aggregation at their isoelectric point because there is no electrostatic repulsion to preventthem from coming close together. In fact, there may even be an electrostatic attractionbetween the positive groups on one protein molecule and the negative groups on another,which promotes protein aggregation and insolubility. A number of proteins are highly solubleacross the whole pH range, because the attractive forces between them are not sufficientlystrong to overcome the thermal energy of the system and so the molecules remain dispersed(Damodaran 1996).

The water solubility of biopolymer molecules is highly dependent on the type and concen-tration of electrolyte ions in solution (Franks 1991). These ions can influence the watersolubility of biopolymers in a number of different ways. At relatively low ionic strengths,electrolyte ions screen electrostatic interactions between molecules (Section 3.4.2). At theisoelectric point, one would expect screening to decrease the strength of the electrostaticattraction between protein molecules and therefore increase solubility. On the other hand, atpH values away from the isoelectric point, one would expect screening to decrease theelectrostatic repulsion between charged biopolymers and therefore decrease their solubility.At intermediate ionic strengths, electrolyte ions may bind to the surface of biopolymers, thusincreasing the short-range hydration repulsion interactions between them and thereby increas-ing their solubility (Israelachvili 1992). Alternatively, the electrolyte may alter the structuralorganization of the water molecules, which can either increase or decrease the magnitude ofthe hydrophobic attraction, depending on the nature of the ions involved (Israelachvili 1992,Damodaran 1996). At high ionic strengths, biopolymer molecules are often precipitated outof solution above a critical salt concentration, which is known as “salting out,” because themajority of water molecules are strongly “bound” to the electrolyte ions and are therefore notavailable to hydrate the biopolymers (Damodaran 1996).

The temperature of an aqueous solution also plays an important role in determining thewater solubility of biopolymers (Damodaran 1996). Altering the temperature of a solutionchanges the relative magnitude of the various kinds of molecular interaction and entropyeffects mentioned in Section 4.6.2. This alters the balance between the forces favoringsolubility and those favoring insolubility. An increase in temperature usually causes anincrease in the strength of the hydrophobic attraction, a decrease in the strength of hydrogenbonds, a decrease in the magnitude of the hydration repulsion, an increase in the strength ofany electrostatic interactions, and an increase in the configurational entropy. The effect oftemperature on the solubility of a biopolymer therefore depends on the relative importanceof the various types of interactions under a given set of experimental conditions.

The water solubility of a biopolymer can be characterized by a solubility index (SI), whichis a measure of the percentage of biopolymer which is soluble in an aqueous solution undera specified set of solvent and environmental conditions (e.g., pH, ionic strength, temperature,and centrifugation speed) (Damodaran 1996). A known concentration of biopolymer isdispersed in solution, and then the solution is ultracentrifuged at a specific speed and time.The concentration of biopolymer remaining in solution is measured using an appropriatetechnique:

SI remaining

initial

= ×100M

M (4.15)

where Mremaining and Minitial are the concentrations of protein in solution after and beforecentrifugation.


The solubility index is often a good indication of the degree of denaturation of a proteiningredient: the lower the solubility index, the greater the degree of denaturation.

4.6.3.2. Emulsification

Biopolymers that have significant amounts of both polar and nonpolar groups tend to besurface active (i.e., they are able to accumulate at oil–water or air–water interfaces (Dickinson1992, Dalgleish 1996a, Damodaran 1996). The major driving force for adsorption is thehydrophobic effect. When the biopolymer is dispersed in an aqueous phase, some of thenonpolar groups are in contact with water, which is thermodynamically unfavorable, be-cause of hydrophobic interactions (Section 4.4.3). When a biopolymer adsorbs to an inter-face, it can adopt a conformation where the nonpolar groups are located in the oil phase(away from the water) and the hydrophilic groups are located in the aqueous phase (incontact with the water). Adsorption also reduces the contact area between the oil and watermolecules at the oil–water interface, which lowers the interfacial tension (Chapter 5). Bothof these factors favor the adsorption of an amphiphilic molecule to an oil–water interface.The conformation that a biopolymer adopts at an interface and the physicochemical prop-erties of the membrane formed depend on its molecular structure and interactions (Das andKinsella 1990, Dickinson 1992, Dalgleish 1996a, Damodaran 1996). Flexible random-coilbiopolymers adopt an arrangement where the predominantly nonpolar segments protrudeinto the oil phase, the predominantly polar segments protrude into the aqueous phase, andthe neutral regions lie flat against the interface (Figure 4.17). The membranes formed bythese types of molecules tend to be relatively open, thick, and of low viscoelasticity.Globular biopolymers (usually proteins) adsorb to an interface so that the predominantlynonpolar regions on the surface of the molecule face the oil phase, while the predominantlypolar regions face the aqueous phase, and so they tend to have a definite orientation at aninterface (Figure 4.18). Once they have adsorbed to an interface, biopolymers often undergostructural rearrangements so that they can maximize the number of contacts between non-polar groups and oil. Random-coil biopolymers are relatively flexible molecules and cantherefore rearrange their structures fairly rapidly, whereas globular biopolymers are morerigid molecules and therefore rearrange more slowly. The unfolding of a globular protein atan interface often exposes amino acids that were originally located in the hydrophobicinterior of the molecule, which can lead to enhanced interactions with neighboring proteinmolecules through hydrophobic attraction or disulfide bond formation (Dickinson andMatsumura 1991, McClements et al. 1993d). Consequently, globular proteins tend to formrelatively thin and compact membranes that have high viscoelasticities (Dickinson 1992).

FIGURE 4.17 The structure of the interfacial membrane depends on the molecular structure andinteractions of the surface-active molecules.


FIGURE 4.18 Extended biopolymers sweep out a large volume of water as they rotate in solution,and so they have a large effective volume fraction.

Membranes formed by globular proteins therefore tend to be more resistant to rupture thanthose formed from random-coil proteins.

To be effective emulsifiers, biopolymers must rapidly adsorb to the surface of the emulsiondroplets created during homogenization and then form a membrane that prevents the dropletsfrom aggregating with one another (Chapter 6). Biopolymer membranes stabilize emulsiondroplets against aggregation by a number of different mechanisms. All biopolymers provideshort-range steric repulsive forces that are usually sufficiently strong to prevent droplets fromgetting close enough together to coalesce (Section 3.5). If the membrane is sufficiently thick,the steric repulsive forces can also prevent droplets from flocculating. Otherwise, it must beelectrically charged so that it can prevent flocculation by electrostatic repulsion (Section 3.4).The properties of emulsions stabilized by charged biopolymers are particularly sensitive topH and ionic strength. At pH values near the isoelectric point of proteins, or at high ionicstrengths, the electrostatic repulsion between droplets may not be sufficiently large to preventthe droplets from aggregating.

A wide variety of proteins are used as emulsifiers in foods because they naturally have ahigh proportion of nonpolar groups and are therefore surface active (Charalambous andDoxstakis 1989, Dickinson 1992, Damodaran 1996). Most polysaccharides are predomi-nantly hydrophilic and are therefore not particularly surface active (BeMiller and Whistler1996). However, a small number of naturally occurring polysaccharides do have somehydrophobic character (e.g., gum arabic) or have been chemically modified to introducenonpolar groups (e.g., some modified starches), and these biopolymers can be used asemulsifiers (BeMiller and Whistler 1996).

4.6.3.3. Thickening and Stabilization

The second major role of biopolymers in food emulsions is to increase the viscosity of theaqueous phase (Mitchell and Ledward 1986). Viscosity enhancement modifies the texture andmouthfeel of food products (“thickening”), as well as reducing the rate at which particlessediment or cream (“stabilization”). Biopolymers used to increase the viscosity of aqueoussolutions are usually highly hydrated and extended molecules or molecular aggregates. Theirability to increase the viscosity of a solution depends principally on their molecular weight,


degree of branching, conformation, and flexibility (Launay et al. 1986, Rha and Pradipasena1986).

The viscosity of a dilute solution of particles increases linearly as the concentration of theparticles increases (Section 8.4):

η η φ= +0 1 2 5( . ) (4.16)

where η is the viscosity of the particulate solution, η0 is the viscosity of the pure solvent,and φ is the volume fraction of particles in solution. The ability of some biopolymers togreatly enhance the viscosity of aqueous solutions when used at very low concentrations isbecause their effective volume fraction is much greater than their actual volume fraction. Abiopolymer rapidly rotates in solution because of its thermal energy, and so it sweeps outa spherical volume of water that has a diameter approximately equal to its end-to-end length(Figure 4.18). The volume of the biopolymer molecule is only a small fraction of the totalvolume of the sphere that is swept out, and so the effective volume fraction of a biopolymeris much greater than its actual volume fraction. The effectiveness of a biopolymer atincreasing the viscosity increases as the volume fraction that it occupies within this spheredecreases. Thus, large highly extended linear biopolymers increase the viscosity moreeffectively than small compact branched biopolymers (Launay et al. 1986, Rha andPradipasena 1986).

In a dilute biopolymer solution, the individual molecules (or molecular aggregates) do notinteract with each other (Dickinson 1992, Lapasin and Pricl 1995). As the concentration ofbiopolymer increases above some critical value (c*), the viscosity of the solution increasesrapidly because the spheres swept out by the biopolymers begin to interact with each other.This type of solution is known as a semidilute solution, because even though the moleculesare interacting with one another, each individual biopolymer is still largely surrounded bysolvent molecules. At still higher polymer concentrations, the molecules pack so closelytogether that they become entangled with each other and the system has more gel-likecharacteristics. Biopolymers which are used to thicken the aqueous phase of emulsions areoften used in the semidilute concentration range (Dickinson 1992).

Solutions containing extended biopolymers often exhibit strong shear-thinning behavior(pseudoplasticity); that is, their apparent viscosity decreases with increasing shear stress(Lapasin and Pricl 1995). Shear thinning occurs because the biopolymer molecules becomealigned with the shear field or because the weak physical interactions responsible for biopoly-mer–biopolymer interactions are disrupted. Some biopolymer solutions have a characteristicyield stress. If such a biopolymer solution experiences an applied stress which is below itsyield stress, it acts like an elastic solid, but when it experiences a stress that exceeds the yieldstress, it acts like a liquid (Section 8.2.3). The characteristic rheological behavior of biopoly-mer solutions plays an important role in determining their functional properties in foodemulsions. For example, a salad dressing must be able to flow when it is poured from acontainer, but must maintain its shape under its own weight after it has been poured onto asalad. The amount and type of biopolymer used must therefore be carefully selected so thatit provides a low viscosity when the salad dressing is poured (high applied stress) but a highviscosity when the salad dressing is allowed to sit under its own weight (low applied stress).The viscosity of biopolymer solutions is also related to the mouthfeel of a food product.Liquids that do not exhibit extensive shear-thinning behavior at the shear stresses experiencedwithin the mouth are perceived as being “slimy.” On the other hand, a certain amount ofviscosity is needed to contribute to the “creaminess” of a product. The shear-thinning behav-ior of biopolymer solutions is also important for determining the stability of food emulsions


to creaming. As an oil droplet moves through an aqueous phase, it only exerts a very smallshear stress on the surrounding liquid. As a result of the shear-thinning behavior of thesolution, it experiences a very high viscosity, which greatly slows down the rate at which itcreams.*

Many biopolymer solutions also exhibit thixotropic behavior (their viscosity decreaseswith time when they are sheared at a constant rate), because the weak physical interactionsthat cause biopolymer molecules to aggregate are disrupted. Once the shearing stress isremoved from the sample, the biopolymer molecules may be able to reform the weak physicalbonds with their neighbors, and so the system regains its original structure and rheologicalproperties. This type of system is said to be reversible, and the speed at which the structureis regained may be important for the practical application of a biopolymer in a food. If thebonds are unable to reform once they are disrupted or if they are only partially reformed, thenthe system is said to be irreversible or partially reversible, respectively.

A food manufacturer must therefore select an appropriate biopolymer or combination ofbiopolymers to produce a final product that has a desirable mouthfeel, stability, and texture.Both proteins and polysaccharides can be used as thickening agents, but polysaccharides areusually preferred because they tend to have higher molecular weights and be more extendedso that they can be used at much lower concentrations.

4.6.3.4. Gelation

Biopolymers are used as functional ingredients in many food emulsions because of theirability to cause the aqueous phase to gel (e.g., yogurts, cheeses, desserts, egg and meatproducts) (Morris 1986, Ledward 1986, Clark and Lee-Tuffnell 1986, Ziegler and Foegedding1990). Gel formation imparts desirable textural and sensory attributes, as well as preventingthe droplets from creaming. A biopolymer gel consists of a three-dimensional network ofaggregated or entangled biopolymers that entraps a large volume of water, giving the wholestructure some “solid-like” characteristics (Mitchell and Ledward 1986, Lapasin and Pricl1995).

The properties of biopolymer gels depend on the type, structure, and interactions of themolecules they contain (Dea 1982, Ziegler and Foegedding 1990). Gels may be transparentor opaque, hard or soft, brittle or rubbery, homogeneous or heterogeneous, and exhibitsyneresis or have good water-holding capacity. Gelation may be induced by a variety ofdifferent methods, including altering the temperature, pH, ionic strength, or solvent qualityor by adding denaturing or cross-linking agents (Ziegler and Foegedding 1990). Biopoly-mers may be cross-linked to one another either by covalent and/or noncovalent bonds. Thetype of cross-links formed depends on the nature of the molecules involved, as well as theprevailing environmental conditions. It is often convenient to distinguish between twodifferent types of gel: particulate and filamentous (Figure 4.19). Particulate gels consist ofbiopolymer aggregates (particles or clumps) which are assembled together to form a three-dimensional network (Doi 1993). This type of gel tends to be formed when the biopolymermolecules are able to interact with their neighbors at any point on their surface. Particulategels are optically opaque because the particles are large enough to strongly scatter light andare prone to syneresis because the relatively large pore sizes between the particles meansthat the water is not held tightly within the gel network by capillary forces. Filamentous gelsconsist of thin filaments of individual or aggregated biopolymer molecules (Doi 1993).Filamentous gels tend to be optically transparent because the filaments are so thin that they

* It should be noted that biopolymers can actually promote creaming at certain concentrations because they causedepletion flocculation (Section 3.6).


do not scatter light significantly. They also tend to have good water-holding capacitybecause the small pore size of the gel network means that the water molecules are heldtightly by capillary forces.

In some foods, a gel is formed on heating (heat-setting gels), while in others it is formedon cooling (cold-setting gels) (Ziegler and Foegedding 1990). Gels may also be eitherthermo-reversible or thermo-irreversible, depending on whether or not gelation is reversible.Gelatin is an example of a cold-setting thermo-reversible gel: when a solution of gelatinmolecules is cooled below a certain temperature, a gel is formed, but when it is reheated, thegel melts (Ledward 1986). Egg white is an example of a heat-setting thermo-irreversible gel:when egg white is heated above a certain temperature, a characteristic white gel is formed,but when it is cooled back to room temperature, it remains as a white gel rather than revertingback into the liquid from which it was formed (Ziegler and Foegedding 1990, Doi 1993).Whether a gel is reversible or irreversible depends on the type of bonds holding the biopoly-mer molecules together, as well as any changes in the molecular structure and organizationof the molecules during gelation. Biopolymer gels that are stabilized by noncovalent inter-actions, and which do not involve large changes in the structure of the individual moleculesprior to gelation, tend to be reversible. On the other hand, gels that are held together bycovalent bonds, or which involve large changes in the structure of the individual moleculesprior to gelation, tend to form irreversible gels.

The type of interaction holding the molecules together in gels varies from biopolymer tobiopolymer. Some proteins and polysaccharides form helical junction zones through exten-sive hydrogen bond formation (e.g., gelatin and gums) (Dea 1982, Ledward 1986, Morris1986). This type of junction zone tends to form when a gel is cooled and be disrupted whenit is heated and is thus responsible for cold-setting gels. Below the gelation temperature, theattractive hydrogen bonds favor junction-zone formation, but above this temperature, theconfigurational entropy favors a random-coil-type structure. Biopolymers with extensivenonpolar groups tend to associate via hydrophobic interactions (e.g., caseins or denaturedwhey proteins) (Ziegler and Foegedding 1990). Electrostatic interactions play an importantrole in determining the gelation behavior of many biopolymers, and so gelation is particu-larly sensitive to the pH and ionic strength of solutions containing these biopolymers(McClements and Keogh 1995). For example, at pH values sufficiently far away fromtheir isoelectric point, proteins may be prevented from gelling because of the strong elec-trostatic repulsion between the molecules; however, if the pH is adjusted near to theisoelectric point, or if salt is added, the proteins tend to aggregate and form a gel (Mulvihilland Kinsella 1987, 1988). The addition of multivalent ions, such as Ca2+, can promotegelation of charged biopolymer molecules by forming salt bridges between the molecules(Morris 1986, Mulvihill and Kinsella 1988). Proteins with thiol groups are capable offorming covalent linkages through thiol–disulfide interchanges, which help to strengthen

FIGURE 4.19 It is often convenient to categorize gels as being either particulate or filamentous,depending on the structural organization of the molecules.


and enhance the stability of gels (Mulvihill and Kinsella 1987). The tendency for a biopoly-mer to form a gel under certain conditions and the physical properties of the gel formeddepend on a delicate balance of various kinds of biopolymer–biopolymer, biopolymer–solvent, and solvent–solvent interactions.

The properties of food emulsions that have a gelled aqueous phase are dependent on thenature of the interactions between the emulsifier adsorbed to the surface of the droplets andthe biopolymer molecules in the gel network (McClements et al. 1993c, Dickinson et al.1996). If there is a strong attractive interaction between the droplet membrane and the gelnetwork, then the network is reinforced and a strong gel is formed. On the other hand, if theemulsifier membrane does not interact favorably with the gel network, then the droplets maydisrupt the network and weaken the gel strength. The magnitude of this effect depends on thesize of the emulsion droplets (McClements et al. 1993c). The larger the droplets comparedto the pore size of the gel network, the greater the disruptive effect. Specific interactionsbetween the proteins and surfactants may also influence the properties of the gels formed(Dickinson and Yamamoto 1996, Dickinson et al. 1996).

4.6.4. Modification of Biopolymers

A wide variety of natural biopolymers could be used as functional ingredients in foods;however, many of these are too difficult to extract economically or only naturally occur insmall quantities. For this reason, only those biopolymers which occur in natural abundance,are easy to extract, and are relatively inexpensive tend to be used as food ingredients. Evenso, these ingredients often do not have the functional properties required for a particularproduct, and so there has been a considerable effort to identify methods of creating newingredients from existing biopolymers.

Food biopolymers can be modified in a variety of ways to improve their functionalproperties (e.g., enzymatically, genetically, chemically, or physically) (Das and Kinsella1990, Dickinson and McClements 1995, Magdassi 1996, Damodaran 1996, BeMiller andWhistler 1996). The molecular structure of a biopolymer can be altered by breaking it downinto smaller subunits, polymerizing it, substituting certain segments of its backbone, co-valently attaching side groups, or physically altering its conformation. Consequently, it ispossible to alter the molecular weight, conformation, charge, hydrophobicity, amphiphilicity,and branching of biopolymer molecules. By carrying out these modifications in a systematicfashion, it is possible to create a range of biopolymers with specific functional properties. Themain challenge for the food scientist is to develop methods of modifying biopolymers thatare both economically viable and legally acceptable.

In addition to their utilization as food ingredients, modified biopolymers have also beenused to study the relationship between the molecular characteristics of biopolymers and theirfunctional properties (Dickinson and McClements 1995). For example, the role of conforma-tion, size, charge, or hydrophobicity can be investigated by altering specific monomers alongthe polymer chain. These studies have provided extremely valuable insights into the molecu-lar basis of the functional properties of biopolymers.

4.6.5. Ingredient Selection

A wide variety of natural and modified biopolymers are available as ingredients in foods,each with its own unique functional properties and optimum range of applications. Foodmanufacturers must decide which biopolymer is the most suitable for each type of foodproduct. The selection of the most appropriate ingredient is often the key to the success ofa particular product. The factors that a manufacturer must consider include the desired


properties of the final product (appearance, rheology, mouthfeel, stability); the composi-tion of the product; the processing, storage, and handling conditions the food will experi-ence during its lifetime; as well as the legal status, cost, availability, consistency frombatch to batch, ease of handling, dispersibility, and functional properties of the biopolymeringredient.


5 Interfacial Properties andTheir Characterization

5.1. INTRODUCTION

The interfacial region which separates the oil from the aqueous phase constitutes only a smallfraction of the total volume of an emulsion (see Table 1.3) Nevertheless, it has a directinfluence on the bulk physicochemical and sensory properties of food emulsions, includingtheir formation, stability, rheology, and flavor. A better understanding of this influence willallow food scientists to create emulsion-based foods with improved quality. In this chapter,we consider the nature of the interfacial region, experimental techniques for characterizingits properties, and the role that it plays in determining the bulk physicochemical propertiesof emulsions.

An interface is a narrow region that separates two phases, which could be a gas and aliquid, a gas and a solid, two immiscible liquids, a liquid and a solid, or two solids. The twophases may consist of different kinds of molecules (e.g., oil and water) or different physicalstates of the same kind of molecule (e.g., liquid water and solid ice). By convention, theregion separating two condensed phases (solids or liquids) is referred to as an interface, whilethe region separating a condensed phase and a gas is called a surface (Everett 1988). Anumber of different types of surfaces and interfaces commonly occur in food emulsions,including oil and water (droplets), air and water (gas cells), ice crystals in water (ice cream),and fat crystals in oil (butter or margarine). In this chapter, the main focus is on the oil–waterinterface because it is present in all food emulsions. Nevertheless, various other types ofsurfaces and interfaces will be considered where appropriate, and it should be recognized thatmuch of the discussion about oil–water interfaces is also applicable to these systems.

5.2. MOLECULAR BASIS OF INTERFACIAL PROPERTIES

5.2.1. Interfaces Between Two Pure Liquids

The interface that separates the oil and water phases is often assumed to be a planar surfaceof infinitesimal thickness (Figure 5.1a). This assumption is convenient for many purposes,but it ignores the highly dynamic nature of the interfacial region, as well as the structure andorganization of the various types of molecules involved (Figure 5.1b). On the molecular level,the oil and water molecules intermingle with each other over distances of the order of a fewmolecular diameters (Everett 1988, Evans and Wennerstrom 1994). The composition of thesystem therefore varies smoothly across the interfacial region (Figure 5.1b), rather thanchanging abruptly (Figure 5.1a). The thickness and dynamics of the interfacial region dependon the relative magnitude of the interactions between the molecules involved (oil–oil, water–water, and water–oil), which can be characterized by the effective interaction parameter (w)discussed in Section 2.6.2 (Evans and Wennerstrom 1994). If this parameter is positive andmuch larger than the thermal energy (w/kT >> 1), the interactions between the different types


of molecules are highly unfavorable and the interfacial region is relatively thin because theprotrusion of a molecule from one phase into another involves a large expenditure of energy.If the effective interaction parameter is approximately equal to the thermal energy (w/kT ≈1), the liquids are partially miscible and the interfacial region increases in thickness. Whenthe effective interaction parameter is negative and much smaller than the thermal energy(w/kT << 1), the two liquids are miscible and the interfacial region disappears altogether. Thethickness and mobility of the interfacial region are therefore governed by a balance betweenthe interaction energies and the thermal energy of the system (Everett 1988, Israelachvili1992).

Oil molecules are incapable of forming hydrogen bonds with water molecules, and so themixing of oil and water is strongly unfavorable because of the hydrophobic effect (Section4.4.3). It is therefore necessary to supply energy to the system in order to increase the contactarea between oil and water molecules. The amount of energy which must be supplied isproportional to the increase in contact area between the oil and water molecules (Hiemenz1986):

∆ ∆G Ai= γ (5.1)

where ∆G is the free energy required to increase the contact area between the two immis-cible liquids by ∆A (at constant temperature and pressure), and γi is a constant of propor-tionality called the interfacial tension. If one of the phases is a gas, the interfacial tensionis replaced by the surface tension (γs). The interfacial tension is a physical characteristic ofa system which is determined by the imbalance of molecular forces across an interface: thegreater the interfacial tension, the greater the imbalance of forces (Israelachvili 1992, Evansand Wennerstrom 1994). Many of the most important macroscopic properties of foodemulsions are governed by this imbalance of molecular forces at an interface, including thetendency for droplets to be spherical, the surface activity of emulsifiers, the nucleation andgrowth of ice and fat crystals, meniscus formation, and the rise of liquids in a capillary tube(Section 5.13).

5.2.2. Interfaces with Adsorbed Emulsifiers

So far, we have only considered the molecular characteristics of an interface that separatestwo pure liquids. In practice, food emulsions contain various types of surface-active mol-

FIGURE 5.1 Interfaces are often assumed to be planar surfaces of infinitesimally small thickness (a),but in reality they are highly dynamic and have a certain thickness which depends on the dimensionsand interactions of the molecules (b).


ecules which can accumulate at the interface and therefore alter its properties (e.g., proteins,polysaccharides, alcohols, and surfactants) (Dickinson and Stainsby 1982, Dickinson 1992).

5.2.2.1. Surface Activity and the Reduction of Interfacial Tension

The surface activity of a molecule is a measure of its ability to accumulate at an interface.A molecule tends to accumulate at an interface when the free energy of the adsorbed stateis significantly lower than that of the unadsorbed state (Hiemenz 1986). The difference in freeenergy between the adsorbed and unadsorbed states (∆Gads) is determined by changes in theinteraction energies of the molecules involved, as well as by various entropy effects (Shaw1980). The change in the interaction energies which occurs as a result of adsorption comesfrom two sources, one associated with the interface and the other with the emulsifier moleculeitself. First, by adsorbing to an oil–water interface, an emulsifier molecule is able to “shield”the oil molecules from the water molecules. The direct contact between oil and watermolecules is replaced by contacts between the nonpolar segments of the emulsifier and oilmolecules and between the polar segments of the emulsifier and water molecules (Israelachvili1992). These interactions are less energetically unfavorable than the direct interactions be-tween oil and water molecules. Second, emulsifier molecules usually have both polar andnonpolar segments, and when they are dispersed in bulk water, some of the nonpolar seg-ments come into contact with water, which is energetically unfavorable because of thehydrophobic effect. By adsorbing to an interface, they are able to maximize the number ofenergetically favorable interactions between the polar segments and water while minimizingthe number of unfavorable interactions between the nonpolar segments and water (Figure5.2). The major driving force favoring the adsorption of an amphiphilic molecule at aninterface is therefore the hydrophobic effect. Nevertheless, various other types of interactionmay also contribute to the surface activity, which may either favor or oppose adsorption (e.g.,hydration repulsion, electrostatic interactions, steric interactions, and hydrogen bonding).

The entropy effects associated with adsorption are mainly due to the fact that when amolecule adsorbs to an interface, it is confined to a region which is considerably smaller thanthe volume it would occupy in a bulk liquid and that its molecular rotation is restricted. Bothof these effects are entropically unfavorable, and so a molecule will only adsorb to aninterface if the energy gained by optimizing the interaction energies is sufficiently large tooffset the entropy lost. When the adsorption energy is large compared to the thermal energy(i.e., –∆Gads >> kT), a molecule “binds” strongly to the surface and has a high surface activity.When the adsorption energy is small compared to the thermal energy (i.e., –∆Gads << kT), amolecule tends to be located mainly in the bulk liquid and has a low surface activity. When

FIGURE 5.2 Surface-active molecules accumulate in the interfacial region because this minimizes thefree energy of the system.


the free energy of adsorption is highly positive (i.e., ∆Gads >> kT), there is a deficit of solutein the interfacial region, which is referred to as negative adsorption (Shaw 1980).

The decrease in the free energy of a system which occurs when a surface-active moleculeadsorbs to an interface manifests itself as a decrease in the interfacial tension (i.e., less energyis required to increase the surface area between the oil and water phases). The extent of thisdecrease depends on the effectiveness of the molecule at “shielding” the direct interactionsbetween the oil and water molecules, as well as on the strength of the interactions betweenthe hydrophilic segments and water, and between the hydrophobic segments and oil(Israelachvili 1992). The more efficient an emulsifier is at shielding the interaction betweenoil and water, the lower the interfacial tension (Figure 5.3).

The ability of surfactant molecules to shield direct interactions between two immiscibleliquids is governed by their optimum packing at an interface, which depends on their molecu-lar geometry (Chapter 4). When the curvature of an interface is equal to the optimumcurvature of a surfactant monolayer (i.e., optimum packing is possible), the interfacial tensionis ultralow because direct interactions between the oil and water molecules are effectivelyeliminated (Section 4.5.3.3). On the other hand, when the curvature of an interface is not atits optimum, the interfacial tension increases because some of the oil molecules are exposedto the polar regions of the surfactant or some of the water molecules come into contact withthe hydrophobic part of the surfactant. Surfactants can usually screen the interactions betweenthe oil and water phases more efficiently than biopolymers, which means they are moreeffective at reducing the interfacial tension. This is because biopolymers cannot pack asefficiently at the interface and because they often have some nonpolar regions on their surfaceexposed to the water phase and some polar regions exposed to the oil phase (Damodaran1996).

The reduction of the interfacial tension by the presence of an emulsifier is referred to asthe surface pressure: π = γo/w – γemulsifier , where γo/w is the interfacial tension of a pure oil–water interface and γemulsifier is the interfacial tension in the presence of the emulsifier (Hiemenz1986).

5.2.2.2. Adsorption Kinetics of Emulsifiers to Interfaces

The tendency of an emulsifier to exist in either the bulk or interfacial regions is governed bythermodynamics; however, the rate at which an emulsifier adsorbs to an interface is deter-mined by various mass transport processes (e.g., diffusion and convection) and energybarriers associated with the adsorption process (e.g., availability of free sites, electrostatic

FIGURE 5.3 The decrease in interfacial tension caused by surface-active molecules depends on howeffectively they “shield” the oil molecules from the water molecules. The interfacial tension of thesystem shown on the left is lower than that on the right, because the contact between the polar andnonpolar molecules is shielded more effectively.


repulsion, steric repulsion, hydrodynamic repulsion, and micelle dynamics). The major fac-tors which influence the adsorption kinetics of an emulsifier at an interface are discussed insome detail in Section 5.7.

5.2.2.3. Conformation of Emulsifiers at Interfaces

The conformation and orientation of molecules at an interface are governed by their attemptto reduce the free energy of the system (Evans and Wennerstrom 1994). Amphiphilic mol-ecules arrange themselves so that the maximum number of nonpolar groups are in contactwith the oil phase, while the maximum number of polar groups are in contact with theaqueous phase (Figure 5.4). For this reason, small-molecule surfactants tend to have theirpolar head groups protruding into the aqueous phase and their hydrocarbon tails protrudinginto the oil phase (Myers 1988). Similarly, biopolymer molecules adsorb so that predomi-nantly nonpolar segments are located within the oil phase, whereas predominantly polarsegments are located within the water phase (Dickinson 1992, Damodaran 1996, Dalgleish1996b). Biopolymer molecules often undergo structural rearrangements after adsorption to aninterface in order to maximize the number of favorable interactions. In aqueous solution,globular proteins adopt a three-dimensional conformation in which the nonpolar amino acidsare predominantly located in the hydrophobic interior of the molecule so that they can beaway from the water (Dill 1990). When they adsorb to an oil–water interface, they are nolonger completely surrounded by water, and so the protein can reduce its free energy byaltering its conformation so that more of the hydrophobic amino acids are located in the oilphase and more of the polar amino acids are located in the water phase (Dalgleish 1996b).The rate at which the conformation of a biopolymer changes at an oil–water interface dependson its molecular structure (Dickinson 1992). Flexible random-coil molecules can rapidly altertheir conformation, whereas rigid globular molecules change more slowly because of variouskinetic constraints. Immediately after adsorption to an interface, a globular protein has aconformation that is similar to that in the bulk aqueous phase. With time, it alters itsconformation so that it can optimize the number of favorable interactions between thenonpolar amino acids and the oil molecules. An intermediate stage in this unfolding processis the exposure of some of the nonpolar amino acids to water, which is energetically unfa-vorable because of the hydrophobic effect, and so there is an energy barrier which must beovercome before unfolding can occur. In this case, the rate of any conformational changeswill depend on the height of the energy barriers compared to the thermal energy.

FIGURE 5.4 The orientation and conformation of molecules at an interface are determined by theirtendency to reduce the free energy of the system.


The configuration of emulsifier molecules at an interface can have an important influenceon the bulk physicochemical properties of food emulsions (Dalgleish 1996b). The coales-cence stability of many oil-in-water emulsions depends on the unfolding and interaction ofprotein molecules at the droplet surface. When globular proteins unfold, they expose reactiveamino acids that are capable of forming hydrophobic and disulfide bonds with their neigh-bors, thus generating a highly viscoelastic membrane that is resistant to coalescence (Dickinsonand Matsumura 1991, Dickinson 1992). The susceptibility of certain proteins to enzymatichydrolysis depends on which side of the adsorbed molecule faces toward the droplet surface(Dalgleish 1996a,b). The susceptibility of surfactants with unsaturated hydrocarbon tails tolipid oxidation depends on whether their tails are perpendicular or parallel to the dropletsurface, the latter being more prone to oxidation by free radicals generated in the aqueousphase.

5.3. THERMODYNAMICS OF INTERFACES

5.3.1. Gas–Liquid Interface in the Absence of an Emulsifier

In the previous section, the relationship between the molecular structure of emulsifiers andtheir surface activity and interfacial conformation was highlighted. In this section, math-ematical quantities and thermodynamic relationships which can be used to describe theproperties of interfaces are introduced. As a whole, emulsions are thermodynamically un-stable systems because of the unfavorable contact between oil and water molecules (Section7.2). Nevertheless, their interfacial properties can often be described by thermodynamicsbecause the adsorption–desorption of emulsifier molecules occurs at a rate which is muchfaster than the time scale of the kinetic destabilization processes (Hunter 1986). Thermody-namically, it is convenient to assume that an interface is a smooth, infinitesimally thin planewhich separates two homogeneous liquids (Figure 5.1a). In order to model a real system, itis necessary to decide exactly where this imaginary plane should be located (Hiemenz 1986).For simplicity, consider a system which consists of liquid water in equilibrium with its vapor(Figure 5.5). The volume fraction of water molecules in the liquid water is approximatelyunity and decreases to approximately zero as one moves up through the interfacial region andinto the vapor phase.

The imaginary plane interface could be located anywhere in the interfacial region indicatedin Figure 5.5 (Hiemenz 1986, Hunter 1986). In practice, it is convenient to assume that theinterface is located at a position where the excess concentration of the substance on one sideof the interface is equal to the deficit concentration of the substance on the other side of theinterface: cexcess = cdeficit. In our example, the excess concentration corresponds to the amountof water above the interface, which is greater than that which would have been present if theconcentration of water was the same as that in the bulk vapor phase right up to the interface.Similarly, the deficit concentration corresponds to the amount of water which is below theinterface, which is lower than that which would have been present if the concentration of thewater was the same as that in the bulk liquid phase right up to the interface. This location ofthe interface is known as the Gibbs dividing surface, after the scientist who first proposed thisconvention.

5.3.2. Gas–Liquid Interface in the Presence of an Emulsifier

The concept of the Gibbs dividing surface is particularly useful for defining the amount ofemulsifier which accumulates at an interface (Hunter 1986). Consider a system whichconsists of a surfactant solution in contact with its vapor (Figure 5.6). The emulsifier isdistributed among the bulk aqueous phase, the vapor, and the interfacial region. The excess


FIGURE 5.6 When an emulsifier is present, the Gibbs dividing surface is conveniently located at theposition where cexcess = cdeficit

for the liquid in which the emulsifier is most soluble, and the surfaceexcess concentration is equal to the shaded region.

FIGURE 5.5 From a thermodynamic standpoint, it is convenient to locate the interface (Gibbsdividing surface) separating a liquid and its vapor where cexcess = cdeficit.

emulsifier concentration at the surface (ni) corresponds to the total amount of emulsifierpresent in the system minus that which would be present if the emulsifier were not surfaceactive and equals the shaded area shown in Figure 5.6. The accumulation of emulsifiermolecules at an interface is characterized by a surface excess concentration (Γ), which isequal to the excess emulsifier concentration divided by the surface area: Γ = ni /A. Foodemulsifiers typically have Γ values of a few milligrams per meter squared (Dickinson 1992,Dalgleish 1996b). It is important to note that the emulsifier molecules are not actuallyconcentrated at the Gibbs dividing surface (which is infinitely thin), because of their finitesize and the possibility of multilayer formation. Nevertheless, this approach is extremelyconvenient for thermodynamic descriptions of the properties of surfaces and interfaces(Hunter 1986). The surface excess concentration is often identified with an experimentallymeasurable parameter called the surface load, which is the amount of emulsifier adsorbedto the surface of emulsion droplets per unit area of interface.


5.3.3. Liquid–Liquid Interface

For an interface between pure oil and pure water, the Gibbs dividing surface could bepositioned at the point where the excess and deficit concentrations of either the oil or thewater were equal on either side of the interface, which will in general be different (Tadrosand Vincent 1983). For convenience, it can be assumed that the phase in which the emulsifieris most soluble is the one used to decide the position of the Gibbs dividing surface. Thesurface excess concentration of the emulsifier is then equal to that which is present in thesystem minus that which would be present if there were no accumulation at the interface(Hunter 1986).

5.3.4. Measurement of the Surface Excess Concentration

The surface excess concentration of an emulsifier can be determined from measurements ofthe variation in the surface tension of an air–liquid interface as the emulsifier concentrationin the bulk liquid is increased (Figure 5.7). There is an equilibrium between emulsifiermolecules at the interface and those in the bulk liquid. As the concentration of emulsifier inthe bulk liquid is increased, so does their concentration at the interface. The presence of theemulsifier molecules at the interface shields the unfavorable contact between the oil andwater molecules and therefore reduces the surface tension. At a certain concentration, thesurface tension reaches a constant value because the surface becomes saturated with emul-sifier molecules. For small-molecule surfactants, the saturation of the surface occurs atapproximately the same concentration as the surfactant molecules form micelles in the bulkliquid (i.e., the critical micelle concentration) (Section 4.5.2). A relationship between thedecrease in surface tension with emulsifier concentration and the amount of emulsifierpresent at the surface can be derived from a mathematical treatment of the thermodynamicsof the system (Hiemenz 1986, Hunter 1986). This relationship is known as the Gibbs isothermequation, which is given by the following expression for an ideal solution:

Γ = −

1RT

d

d x

γln( )

for nonionic emulsifiers (5.2)

Γ = −

12RT

d

d x

γln( )

for ionic emulsifiers (5.3)

where x is the concentration of emulsifier in the aqueous phase, R is the gas constant, and Tis the absolute temperature. These equations are used to determine the surface excess concen-tration of emulsifiers from experimental measurements of the surface tension versus emulsi-fier concentration, with Γ as the slope of the initial part of the curve (Figure 5.7). The factor2 appears in the denominator of Equation 5.3 because the counterions associated with thehead groups of ionic surfactants also accumulate at the interface (Hunter 1986). Equation 5.3is only applicable at low ionic strengths where the interaction between the head groups andcounterions is strong. As the ionic strength increases, the electrostatic interactions betweenthe head groups and counterions are screened, and so Equation 5.2 becomes more applicable.Knowledge of the surface excess concentration is important for formulating food emulsionsbecause it determines the minimum amount of emulsifier which can be used to create anemulsion with a given size distribution. The smaller the value of Γ, the greater the area ofoil–water interface which can be covered per gram of emulsifier, and therefore the smaller


FIGURE 5.7 Dependence of the surface tension on the concentration of emulsifier. This type of plotcan be used to measure the critical micelle concentration (CMC), the surface excess concentration, andmaximum surface pressure of emulsifiers.

the size of droplets which can be effectively stabilized by the emulsifier. Plots of surfacetension versus emulsifier concentration are also useful because they indicate the maximumsurface pressure (πmax) which can be achieved when the surface is saturated by an emulsifier,which has important consequences for the formation and stability of food emulsions (Chap-ters 6 and 7).

5.4. PROPERTIES OF CURVED INTERFACES

The majority of surfaces or interfaces found in food emulsions are curved rather than planar.The curvature of an interface alters its characteristics in a number of ways. The interfacialtension tends to cause an emulsion droplet to shrink in size so as to reduce the unfavorablecontact area between the oil and water phases (Hunter 1986, Everett 1988). As the dropletshrinks, there is an increase in its internal pressure because of the compression of the watermolecules. Eventually, an equilibrium is reached where the inward stress due to the interfacialtension is balanced by the outward stress associated with compressing the bonds between theliquid molecules inside the droplet.* At equilibrium, the pressure within the droplet is largerthan that outside and can be related to the interfacial tension and radius of the droplets usingthe Laplace equation (Atkins 1994):

∆pr

= 2γ(5.4)

This equation indicates that the pressure difference across the surface of an emulsiondroplet increases as the interfacial tension increases or the size of the droplet decreases. Theproperties of a material depend on the pressure exerted on it, and so the properties of a

* The shrinkage of a droplet due to the interfacial tension is usually negligibly small, because liquids have a verylow compressibility.


material within a droplet are different from those of the same material in bulk (Atkins 1994).This effect is usually negligible for liquids and solids which are contained within particlesthat have radii greater than a few micrometers, but it does become significant for smallerparticles (Hunter 1986). For example, the water solubility of oil increases as the radius of anoil droplet decreases (Dickinson 1992, Atkins 1994):

SS

v

rRT*=

exp

2γ(5.5)

where S is the water solubility of the oil in the droplet, S* is the water solubility of bulkoil, and v is the molar volume of the oil. For a typical food oil (v = 10–3 m3 mol–1, γ = 10 mJm–2), the value of S/S* is 2.24, 1.08, 1.01, and 1.0 for oil droplets with radii of 0.01, 0.1, 1,and 10 µm, respectively. Equation 5.5 has important implications for the stability of emulsiondroplets, fat crystals, and ice crystals to Ostwald ripening (Chapter 7).

So far, we have assumed that the interfacial tension of a droplet is independent of its radius.Experimental work has indicated that this assumption is valid for oil droplets, even down tosizes where they only contain a few molecules, but that it is invalid for water droplets belowa few nanometers because of the disruption of long-range hydrogen bonds (Israelachvili1992).

Emulsifier molecules have a major influence on the properties of curved surfaces andinterfaces. Each type of surfactant has an optimum curvature which is governed by itsmolecular geometry and interactions with its neighbors (Section 4.5.3). When the curvatureof an interface is equal to the optimum curvature of a surfactant monolayer, the interfacialtension is extremely low because the interactions between the emulsifier molecules areoptimized and shielding between the oil and water molecules is extremely efficient. On theother hand, when the curvature of the interface is not close to the optimum curvature of thesurfactant monolayer, the interfacial tension increases because the interactions between thesurfactant molecules are not optimum and the shielding between oil and water molecules isless efficient. The interfacial tension of an interface therefore depends on the moleculargeometry of the surfactant used.

5.5. CONTACT ANGLES AND WETTING

In food systems, we are often interested in the ability of a liquid to spread over or “wet” thesurface of another material. In some situations, it is desirable for a liquid to spread over asurface (e.g., when coating a food with an edible film), while in other situations it is importantthat a liquid does not spread (e.g., when designing waterproof packaging). When a drop ofliquid is placed on the surface of a material, it may behave in a number of ways, dependingon the nature of the interactions between the various types of molecules present. The twoextremes of behavior that are observed experimentally are outlined below (Figure 5.8):

1. Poor wetting. The liquid gathers up into a lens, rather than spreading across thesurface of a material.

2. Good wetting. The liquid spreads over the surface of the material to form a thinfilm, which has a liquid–gas interface and a liquid–solid interface.

The situation that occurs in practice depends on the relative magnitude of the interactionsbetween the various types of molecules involved (i.e., solid–liquid, solid–gas, and liquid–gas). A system tends to organize itself so that it can maximize the number of favorable


FIGURE 5.8 The wetting of a surface by a liquid depends on a delicate balance of molecularinteractions among solid, liquid, and gas phases.

interactions and minimize the number of unfavorable interactions between the molecules.Consider what may happen when a drop of liquid is placed on a solid surface (Figure 5.8).If the liquid remained as a lens, there would be three different interfaces: solid–liquid, solid–gas, and liquid–gas, each with its own interfacial or surface tension. If the liquid spread overthe surface, there would be a decrease in the area of the solid–gas interface but an increasein the areas of both the liquid–gas and solid–liquid interfaces. The tendency for a liquid tospread therefore depends on the magnitude of the solid–gas interactions (γSG) compared to themagnitude of the solid–liquid and liquid–gas interactions that replace it (γSL + γLG). Thissituation is conveniently described by a spreading coefficient, which is defined as (Hunter1993):

S = − +γ γ γSG SL LG( ) (5.6)

If the energy associated with the solid–gas interface is greater than the sum of the energiesassociated with the solid–liquid and liquid–gas interfaces (γSG > γSL + γLG), then S is positiveand the liquid tends to spread over the surface to reduce the energetically unfavorable contactarea between the solid and the gas. On the other hand, if the energy associated with the solid–gas interface is less than that associated with forming the solid–liquid and liquid–gas inter-faces (γSG < γSL + γLG), then S is negative and the liquid tends to form a lens.

The shape of a droplet can be predicted by carrying out a force balance at the point on thesurface where the solid, liquid, and gas meet (Figure 5.9) using the Young equation (Hiemenz1986):

γ γ γ θSG SL LG= + cos (5.7)

so that:


cosθγ γ

γ=

−SG SL

LG(5.8)

Here, θ is known as the contact angle, which is the angle of a tangent drawn at the point wherethe liquid contacts the surface. By convention, this angle is measured from the side of thedroplet (Figure 5.9). The shape of a droplet on a surface can therefore be predicted from aknowledge of the contact angle: the smaller θ, the greater the tendency for the liquid to spreadover the surface.

So far, we have only considered the situation where a liquid spreads over a solid surface,but similar equations can be used to consider other three-component systems, such as a liquidspreading over the surface of another liquid (e.g., oil, water, and air) or a crystal at aninterface between two other liquids (e.g., a fat crystal at an oil–water interface). The lattercase is important when considering the nucleation and location of fat crystals in oil dropletsand has a pronounced influence on the stability and rheology of many important foodemulsions, including milk, cream, butter, and whipped cream (Walstra 1987, Boode 1992,Dickinson and McClements 1995).

The above equations assume that the materials involved are completely insoluble in eachother, so that the values of γSG, γSL, and γLG (or the equivalent terms for other three-componentsystems) are the same as those for pure systems. If the materials are partially miscible, thenthe interfacial tensions will change over time until equilibrium is reached (Hunter 1993). Thesolubility of one component in another generally leads to a decrease in the interfacial tension.This means that the shape that a droplet adopts on a surface may change with time (e.g., aspread liquid may gather into a lens, or vice versa, depending on the magnitude of the changesin the various surface or interfacial tensions).

The contact angle of a liquid can conveniently be measured using a microscope, which isoften attached to a computer with video image analysis software (Hunter 1986). A droplet ofthe liquid to be analyzed is placed on a surface and its shape is recorded via the microscope.The contact angle is determined by analyzing the shape of the droplet. The advantages anddisadvantages of a variety of other techniques available for measuring contact angles havebeen considered by Hunter (1986).

The concepts of a contact angle and a spreading coefficient are useful for explaining anumber of important phenomena which occur in food emulsions. The contact angle deter-mines the distance that a fat crystal protrudes from the surface of a droplet into the surround-ing water (Boode 1992) and whether nucleation occurs within the interior of a droplet or atthe oil–water interface (Dickinson and McClements 1995). It also determines the amount ofliquid which is drawn into a capillary tube and the shape of the meniscus at the top of theliquid (Hunter 1986, Hiemenz 1986). A knowledge of the contact angle is also often requiredin order to make an accurate measurement of the surface or interfacial tension of a liquid(Couper 1993).

FIGURE 5.9 Force balance of a droplet at a solid–gas interface.


5.6. CAPILLARY RISE AND MENISCUS FORMATION

The surface tension of a liquid governs the rise of liquids in capillary tubes and the formationof menisci (curved surfaces) at the top of liquids (Hiemenz 1986, Hunter 1986). When a glasscapillary tube is dipped into a beaker of water, the liquid climbs up the tube and forms acurved surface (Figure 5.10). The origin of this phenomenon is the imbalance of intermolecu-lar forces at the various surfaces and interfaces in the system (Evans and Wennerstrom 1994).When water climbs up the capillary tube, some of the air–glass contact area is replaced bywater–glass contact area, while the air–water contact area remains fairly constant. This occursbecause the imbalance of molecular interactions between glass and air is much greater thanthat between glass and water. Consequently, the system attempts to maximize the glass–watercontacts and minimize the glass–air contacts by having the liquid climb up the inner surfaceof the capillary tube. This process is opposed by the downward gravitational pull of the liquid.When the liquid has climbed to a certain height, the surface energy it gains by optimizing thenumber of favorable water–glass interactions is exactly balanced by the potential energy thatmust be expended to raise the mass of water up the tube (Hiemenz 1986). A mathematicalanalysis of this equilibrium leads to the derivation of the following equation:

γρ

θ=

∆ ghr

2 cos(5.9)

where g is the gravitational constant, h is the height that the meniscus rises above the levelof the water, r is the radius of the capillary tube, ∆ρ is the difference in density between thewater and the air, and θ is the contact angle (Figure 5.10). This equation indicates that thesurface tension can be estimated from a measurement of the height that a liquid rises up acapillary tube and the contact angle: the greater the surface tension, the higher the liquid risesup the tube. This is one of the oldest and simplest methods of determining the surface tensionof pure liquids, but it has a number of problems which limit its application to emulsifiersolutions (Couper 1993).

Capillary forces are responsible for the entrapment of water in biopolymer networks andoil in fat crystal networks. When the gaps between the network are small, the capillary forceis strong enough to hold relatively large volumes of liquid, but when the gaps exceed acertain size, the capillary forces are no longer strong enough and syneresis or “oiling off ”

FIGURE 5.10 The rise of a liquid up a capillary tube is a result of its surface tension.


occurs. A knowledge of the origin of capillary forces is therefore important for an under-standing of the relationship between the microstructure of foods and many of their qualityattributes.

5.7. ADSORPTION KINETICS

The rate at which an emulsifier adsorbs to an interface is one of the most important factorsdetermining its efficacy as a food ingredient (Magdassi and Kamyshny 1996, Walstra 1996b).The adsorption rate depends on the molecular characteristics of the emulsifier (e.g., size,conformation, and interactions), the nature of the bulk liquid (e.g., viscosity), and the prevail-ing environmental conditions (e.g., temperature and mechanical agitation). It is often conve-nient to divide the adsorption process into two stages: (1) movement of the emulsifiermolecules from the bulk liquid to the interface and (2) incorporation of the emulsifiermolecules at the interface. In practice, emulsifier molecules are often in a dynamic equilib-rium between the adsorbed and unadsorbed states, and so the rate at which emulsifiermolecules leave the interface must also be considered (Hunter 1993, Magdassi and Kamyshny1996).

5.7.1. Movement of Molecules to an Interface

In this section, we assume that an emulsifier molecule is adsorbed to an interface as soon asit encounters it (i.e., there are no energy barriers that retard adsorption). In an isothermalquiescent liquid, emulsifier molecules move from a bulk liquid to an interface by moleculardiffusion, with an initial adsorption rate given by (Tadros and Vincent 1983, Magdassi andKamyshny 1996):

ddt

cDt

Γ =π (5.10)

where D is the translational diffusion coefficient, Γ is the surface excess concentration, t isthe time, and c is the concentration of emulsifier initially in the bulk liquid. The variation ofthe surface excess concentration with time is obtained by integrating this equation withrespect to time:

Γ ( )t cDt= 2π (5.11)

Thus, a plot of the surface excess concentration versus √t should be a straight line thatpasses through the origin. This equation indicates that the accumulation of an emulsifier atan interface occurs more rapidly as the concentration of emulsifier in the bulk liquidincreases or as the diffusion coefficient of the emulsifier increases. The diffusion coefficientincreases as the size of molecules decreases, and one would therefore expect smallermolecules to adsorb more rapidly than larger ones. Experiments with proteins have shownthat Equation 5.11 gives a good description of the early stages of adsorption to cleaninterfaces (Damodaran 1989, Walstra 1996b). After the initial stages, the adsorption ratedecreases because the interface becomes saturated with emulsifier molecules and thereforethere are fewer sites available for the emulsifier to adsorb to (Figure 5.11). In practice, therate may be faster than that given by Equation 5.10 because of convection currents causedby temperature gradients within a liquid. Consequently, considerable care must be taken to


ensure that the temperature within a sample is uniform when measuring diffusion-controlledadsorption processes.

The above equations do not apply during the homogenization of emulsions, becausehomogenization is a highly dynamic process and mass transport is governed mainly byconvection rather than diffusion (Dickinson 1992, Walstra 1996b). Under isotropic turbulentconditions, the initial increase of the surface excess concentration with time is given by(Dukhin et al. 1995):

Γ ( )t Cr cr

rtd

e

d

= +

13

(5.12)

where C is a constant which depends on the experimental conditions, and rd and re are theradii of the droplet and emulsifier, respectively. This equation predicts that the adsorption rateincreases as the concentration of emulsifier increases, the size of the emulsion dropletsincreases, or the size of the emulsifier molecules increases relative to the size of the droplets.This equation implies that when an emulsion is homogenized, the emulsifier moleculesinitially adsorb preferentially to the larger droplets and that larger emulsifier moleculesadsorb more rapidly than smaller ones (which is the opposite of diffusion-controlled adsorp-tion). This explains why large casein micelles adsorb faster than individual casein moleculesduring the homogenization of milk (Mulder and Walstra 1974).

5.7.2. Incorporation of Emulsifier Molecules at an Interface

So far, we have assumed that as soon as an emulsifier molecule reaches an interface, it isimmediately adsorbed. In practice, there may be one or more energy barriers that must beovercome before a molecule adsorbs, and so only a fraction of the encounters between anemulsifier molecule and an interface lead to adsorption (Damodaran 1996, Magdassi andKamyshny 1996). In these systems, adsorption kinetics may be governed by the height of theenergy barrier, rather than by the rate at which the molecules reach the interface (Magdassiand Kamyshny 1996).

There are a number of reasons why an energy barrier to adsorption may exist:

1. As an emulsifier molecule approaches an interface, there may be various types ofrepulsion interactions between it and the emulsifier molecules already adsorbedto the interface (e.g., electrostatic, steric, hydration, or thermal fluctuation) (Chap-ter 3).

2. Some surface-active molecules will only be adsorbed if they are in a specificorientation when they encounter the interface. For example, it has been suggested

FIGURE 5.11 Adsorption kinetics for a diffusion-controlled system.


that globular proteins which have hydrophobic patches on their surface must facetoward an oil droplet during an encounter (Damodaran 1996).

3. The ability of surfactant molecules to form micelles plays a major role in determin-ing their adsorption kinetics (Dukhin et al. 1995, Stang et al. 1994, Karbstein andSchubert 1995). Surfactant monomers are surface active because they have a polarhead group and a nonpolar tail, but micelles are not surface active because theirexterior is surrounded by hydrophilic head groups. The adsorption kinetics there-fore depends on the concentration of monomers and micelles present, as well as thedynamics of micelle formation–disruption.

Micelle dynamics can be characterized by two relaxation times, one associated with themovement of individual monomers in and out of the micelles (τfast) and the other associatedwith the complete dissolution of a whole micelle into a number of individual molecules (τslow)(Land and Zana 1987). Consider the processes that occur when a fresh oil–water interface isbrought into contact with an aqueous solution containing surfactant monomers and micelles(Kabalnov and Weers 1996). When a monomer adsorbs to the surface, the local monomerconcentration is depleted, and so there is a concentration gradient between the region in theimmediate vicinity of the interface and the bulk solution. This gradient can be equalized bythe diffusion of surfactant monomers from the bulk liquid into the depleted zone. However,this disturbs the equilibrium between the monomers and micelles, leading to the partialdissociation of the micelles. Thus the micelles influence the adsorption kinetics indirectlyrather than directly. The adsorption kinetics therefore depends on the diffusion coefficient ofthe monomers and micelles, as well as the dynamic properties of the micelles (i.e., therelaxation times). When the micelle relaxation mechanisms that release the monomers aremuch slower than the diffusion of the monomers, the adsorption kinetics is governed mainlyby the critical micelle concentration, rather than the actual micelle concentration present. Onthe other hand, if the micelle relaxation mechanisms occur rapidly, then the micelles act asan additional source of monomers, and the adsorption rate is enhanced when the surfactantconcentration is increased above the critical micelle concentration.

The adsorption of emulsifier molecules at a surface or interface can be measured using avariety of experimental methods (Couper 1993, Kallay et al. 1993, Dukhin et al. 1995). Themost commonly used method is to measure the variation in surface or interfacial tension withtime using a tensiometer (Section 5.9). A number of workers have also used radiolabeledemulsifier molecules to measure adsorption kinetics (Damodaran 1989). A radioactivitydetector is placed immediately above a water–air surface. The radiolabeled emulsifier isinjected into the water and the increase in the radioactivity at the surface is recorded over timeby the detector. The radioactivity is highly attenuated by the water, and so only thosemolecules which are close to the air–water surface are detected. A variety of other experimen-tal methods that can detect changes in interfacial properties due to the adsorption of emul-sifier molecules have also been used to monitor adsorption kinetics, including ultraviolet-visible spectroscopy, infrared spectroscopy, fluorescence spectroscopy, ellipsometry, andinterfacial rheology (Kallay et al. 1993).

5.8. COMMON TYPES OF INTERFACIAL MEMBRANES

A variety of different types of interfacial membranes are commonly found in food emulsionsas a result of the different types of surface-active materials in the system. The four mostcommon types of interface are discussed briefly below:

1. Solid particles. Many foods contain tiny solid particles that are capable of adsorbingto the surface of emulsion droplets (e.g., mustard powder and spices) (Dickinson


and Stainsby 1982, Dickinson 1992). The arrangement of these particles at aninterface depends on whether they are preferentially wetted by the continuous orthe dispersed phase. Those particles which are wetted best by the dispersed phaseprotrude into the droplet, whereas those which are wetted best by the continuousphase protrude into the surrounding liquid. Obviously, those particles which pro-trude into the continuous phase are most suitable for providing protection againstdroplet aggregation.

2. Flexible biopolymers. A number of surface-active proteins and polysaccharideshave fairly flexible random-coil-type structures (Section 4.6). These biopolymerstend to form relatively thick interfacial membranes with fairly open structures. Tobe effective at preventing droplet aggregation, part of the molecules must have ahigh affinity for the continuous phase and extend a significant distance into it.

3. Globular biopolymers. Many surface-active proteins have a compact globularstructure and form relatively thin but dense interfacial layers. These proteinsusually provide protection against aggregation by a combination of electrostaticand steric interactions (Chapter 3).

4. Small-molecule surfactants. Surfactants form a fairly dense interfacial mem-brane which prevents molecular contact between the liquid in the droplets and thatin the continuous phase. At low surfactant concentrations, this membrane is usuallya monolayer, but at high concentrations, it may be a multilayer (Friberg and El-Nokaly 1983).

5.9. INTERFACIAL COMPOSITION AND COMPETITIVE ADSORPTION

Most food emulsions contain droplets which are coated by a mixture of different types ofsurface-active components (e.g., proteins, polysaccharides, phospholipids, and surfactants),rather than a single chemically pure type (Dickinson 1992; Dalgleish 1996a,b; Walstra1996b). Bulk physicochemical properties of emulsions, such as their ease of formation,stability, and texture, are governed by the nature of the interface, and therefore it is importantfor food scientists to understand the factors which determine the composition of the interfa-cial region (Dalgleish 1996a). Interfacial composition is determined by the type and concen-tration of surface-active components present, their relative affinity for the interface, themethod used to prepare the emulsion, solution conditions (e.g., temperature, pH, and ionicstrength), and the history of the emulsion. In this section, some of the most important factorswhich influence the interfacial composition of food emulsions are reviewed.

Interfacial composition depends on the relative adsorption rates of the various types ofsurface-active components which are present during the homogenization process, as well asany changes that may occur after homogenization (Dickinson 1992). Immediately afterhomogenization, the droplets tend to be coated by those surface-active molecules whichabsorb to the interface most rapidly under turbulent conditions (Section 5.7). Nevertheless,the interfacial composition may change with time after homogenization because some of thesurface-active molecules that were initially present at the droplet surface are displaced bymolecules in the bulk liquid that have a greater affinity for the surface. Alternatively,additional surface-active components may be added to the continuous phase of an emulsionafter homogenization, and these may displace some of the emulsifier molecules at the dropletsurface. The displacement of emulsifier molecules from an interface may be retarded if theyare capable of undergoing some form of conformational change that enables them to bindstrongly to their neighbors. For example, when an emulsion is heated under alkaline condi-tions, β-lactoglobulin is capable of unfolding and forming extensive covalent (disulfide)bonds with its neighbors, which makes it difficult to displace from an interface, even by moresurface-active molecules (McClements et al. 1993d). The factors that determine the rate at


FIGURE 5.12 Comparison of the affinity of amphiphilic biopolymers and small-molecule surfactantsfor an oil–water interface.

which emulsifier molecules are adsorbed to an interface are discussed in Section 5.7. In thissection, the factors which determine the relative affinity of surface-active molecules for aninterface are considered.

The affinity of an emulsifier molecule for an interface can be described by its adsorptionefficiency and its surface activity (Dickinson 1992). The adsorption efficiency is a measureof the minimum amount of emulsifier required to saturate an interface, whereas the surfaceactivity is a measure of the maximum decrease in interfacial tension achievable when aninterface is completely saturated (i.e., Πmax). Adsorption efficiencies and surface activitiesdepend on the molecular structure of emulsifiers, as well as the prevailing environmentalconditions. Amphiphilic biopolymers, such as proteins, tend to have higher adsorption effi-ciencies, but lower surface activities, than small-molecule surfactants (Figure 5.12).

Emulsifier molecules in solution partition themselves between the bulk solution and theinterfacial region. The equilibrium constant between the adsorbed and unadsorbed states isproportional to eE/kT, where E is a binding energy (Hiemenz 1986, Dickinson 1992). A smallmolecule tends to have one fairly strong binding site (its hydrophobic tail), whereas biopoly-mer molecules tend to have a large number of relatively weak binding sites (nonpolar aminoacid side groups). The overall binding energy of a biopolymer molecule tends to be greaterthan that of a small-molecule surfactant, and therefore it binds more efficiently (i.e., lessemulsifier must be added to the bulk aqueous phase before the interface becomes completelysaturated). For the same reason, small-molecule surfactants tend to rapidly exchange betweenthe adsorbed and unadsorbed states, whereas biopolymer molecules tend to remain at aninterface for extended periods after adsorption. The relatively slow desorption of biopolymermolecules from an interface compared to small-molecule surfactants is best illustrated by ananalogy. Consider a flock of birds that are all connected to each other by a piece of string.At any particular time, an individual bird may decide to fly, but it cannot get very far fromthe ground because it is held back by the other birds. It is only when a large number of birdssimultaneously decide to take off together that they are all able to leave the ground, whichis statistically very unlikely. Thus, at low emulsifier concentrations, biopolymers have agreater affinity for an interface than small-molecule surfactants.

On the other hand, small-molecule surfactants tend to decrease the interfacial tension bya greater amount than biopolymer molecules at concentrations where the interface is com-pletely saturated, because they pack more efficiently and therefore screen the unfavorableinteractions between the oil and water molecules more effectively. Thus, at high emulsifierconcentrations, small-molecule surfactants have a greater affinity for an interface than biopoly-


mers and will tend to displace them. This accounts for the ability of relatively high concen-trations of surfactant molecules (e.g., 1% Tween 20) to displace proteins from the surface ofoil droplets (Dickinson et al. 1993c; Courthaudon et al. 1991a,b,c,d; Dickinson and Tanai1992; Dickinson 1992; Dickinson and Iveson 1993). Small-molecule surfactants are oftenadded to ice cream premixes so as to displace the proteins from the surface of the milk fatglobules prior to cooling and shearing (Goff et al. 1987). This causes the droplets to becomemore susceptible to partial coalescence, which leads to the formation of a network of aggre-gated droplets which stabilizes the air bubbles and gives the final product its characteristicshelf life (Berger 1976).

The ease with which proteins can be displaced from an oil–water interface often dependson the age of the interfacial membrane. Some globular proteins become surface denaturedafter adsorption to an interface because of the change in their environment (Dickinson andMatsumura 1991, McClements et al. 1993d). When the proteins unfold, they expose aminoacids that are capable of forming disulfide bonds with their neighbors and thus form aninterfacial membrane that is partly stabilized by covalent bonds. This accounts for theexperimental observation that the ease with which β-lactoglobulin can be displaced from thesurface of oil droplets decreases as the emulsion ages (Dalgleish 1996a).

The phase in which a surfactant is most soluble also determines its effectiveness atdisplacing proteins from an interface. For example, water-soluble surfactants have beenshown to be more effective at displacing proteins from the surface of oil droplets than oil-soluble surfactants (Dickinson et al. 1993a,b).

Interfacial composition also depends on the relative concentration of the different types ofemulsifiers present. A minor surface-active ingredient may make up a substantial fraction ofthe droplet membrane when the droplets are large (small total surface area), but have anegligible contribution when the droplets are small (large total surface area).

The thermal history of an emulsion often has an important influence on the compositionof its interfacial layer. This is particularly true for globular proteins that are capable ofundergoing conformational changes at elevated temperatures. The interfacial concentration ofβ-lactoglobulin in oil-in-water emulsions increases when they are heated to 70°C (Dickinsonand Hong 1994). In addition, the protein becomes more difficult to displace from the interfaceusing small-molecule surfactants. The most likely explanation for this behavior is the thermaldenaturation of the β-lactoglobulin molecules at this temperature. Thermal denaturationcauses the molecules to unfold and expose amino acids that were originally located in theirinterior, including amino acids with nonpolar and cysteine side groups. The increase insurface hydrophobicity of the protein enhances its affinity for the droplet surface, as well asfor other protein molecules, which accounts for the increase in interfacial concentration. Theability of the protein to form disulfide bonds with its neighbors leads to a covalently bondedfilm, which accounts for the difficulty in displacing the protein with surfactants. Temperaturehas been found to influence the ability of small-molecule surfactants to displace β-caseinfrom the surface of oil droplets (Dickinson and Tanai 1992). A minimum in the interfacialconcentration of protein was observed at temperatures between 5 and 10°C.

The pH and ionic strength of the aqueous phase containing the emulsifier molecules alsohave an important influence on adsorption kinetics and interfacial composition (Hunt andDalgleish 1994, 1996). Potassium chloride has been shown to affect the competition betweenproteins for an interface (Hunt and Dalgleish 1995). Increasing the concentration of KClpresent in the aqueous phase prior to homogenization causes a decrease in β-lactoglobulinand an increase in α-lactalbumin at the interface of oil-in-water emulsions at pH 7, but hasno influence on interfacial composition for emulsions at pH 3. Changes in the relativeproportions of the different types of caseins were also observed for emulsions stabilized bycaseinate when the KCl concentration was increased. Potassium chloride was also found to


alter the competitive adsorption of proteins added after homogenization (Hunt and Dalgleish1996).

The above discussion highlights the wide variety of factors which can influence interfacialcomposition, such as the emulsifier concentration, emulsifier type, solution conditions, tem-perature, and time. For this reason, a great deal of research is being carried out to establishthe relative importance of each of these factors and to establish the relationship betweeninterfacial composition and the bulk physicochemical properties of food emulsions.

5.10. MEASUREMENT OF SURFACE AND INTERFACIAL TENSIONS

By definition, surface tension is measured at a gas–fluid interface (e.g., air–water or oil–water), while interfacial tension is measured at a fluid–fluid interface (e.g., oil–water). Thesequantities are measured using instruments called surface or interfacial tensiometers, respec-tively (Couper 1993). Tensiometers can be used to provide valuable information about thecharacteristics of surfaces and interfaces and about the properties of emulsifiers in solution,such as the surface excess concentration, surface pressure, critical micelle concentration, andadsorption kinetics (Hiemenz 1986, Hunter 1986, Evans and Wennerstrom 1994). A widevariety of different types of tensiometer are available for providing information about theproperties of surfaces and interfaces (Couper 1993). These instruments differ according to thephysical principles on which they operate, their mechanical design, whether measurementsare static or dynamic, and whether they are capable of measuring surface tension, interfacialtension, or both. Static measurements are carried out on surfaces or interfaces which areconsidered to be at equilibrium, whereas dynamic measurements are carried out on surfacesor interfaces which are not at equilibrium and can therefore be used to monitor adsorptionkinetics.

5.10.1. Du Nouy Ring Method

The Du Nouy ring method is used to measure static surface and interfacial tensions of liquids(Hiemenz 1986, Couper 1993). The apparatus required to carry out this type of measurementconsists of a vessel containing the liquid(s) to be analyzed and a ring which is attached toa sensitive force-measuring device (Figure 5.13). The vessel is capable of being moved

TABLE 5.1Summary of the Instruments Used for MeasuringSurface and Interfacial Tensions

Surface tension Interfacial tension

EquilibriumDu Nouy ring Du Nouy ringWilhelmy plate Wilhelmy platePendant drop Pendant dropSessile drop Sessile dropSpinning drop Spinning dropCapillary rise

DynamicMaximum bubble pressure Drop volumeOscillating jet Pendant dropDrop volume Spinning dropSurface waves


FIGURE 5.13 Du Nouy ring method of determining the interfacial and surface tension of liquids.

upward and downward in a controlled manner while the position of the ring is kept constant.Initially, the vessel is positioned so that the ring is submerged just below the surface of theliquid being analyzed. It is then slowly lowered and the force exerted on the ring is recorded.As the surface of the liquid moves downward, some of the liquid “clings” to the ring becauseof its surface tension (Figure 5.13). The weight of the liquid that clings to the ring isrecorded by the force-measuring device and is related to the force that results from thesurface tension.

The Du Nouy ring method is usually used in a “detachment” mode. The vessel is lowereduntil the liquid clinging to the ring ruptures and the ring becomes detached from the liquid.The force exerted on the ring at detachment is approximately equal to the surface tensionmultiplied by the length of the ring perimeter: F = 4πRγ, where R is the radius of the ring.In practice, this force has to be corrected because the surface tension does not completely actin the vertical direction and because some of the liquid remains clinging to the ring after ithas become detached.

F R= 4π γβ (5.13)

where β is a correction factor which depends on the dimensions of the ring and the densityof the liquid(s) involved. Values of β have been tabulated in the literature or can becalculated using semiempirical equations (Couper 1993). One of the major problems asso-ciated with the Du Nouy ring method, as well as any other detachment method, is thatserious errors may occur when measuring the surface tension of emulsifier solutions ratherthan pure liquids. When a ring detaches from a liquid, it leaves behind some fresh surfacewhich is initially devoid of emulsifier. The measured surface tension therefore depends onthe speed at which the emulsifier molecules diffuse from the bulk liquid to the fresh surfaceduring the detachment process. If an emulsifier adsorbs rapidly compared to the detachmentprocess, the surface tension measured by the detachment method will be the same as theequilibrium value, but if the emulsifier adsorbs relatively slowly, the surface tension will begreater than the equilibrium value because the surface has a lower emulsifier concentrationthan expected.

The Du Nouy ring method can also be used to determine the interfacial tension betweentwo liquids (Couper 1993). In this case, the ring is initially placed below the surface of themost dense liquid (usually water). The force acting on the ring is then measured as it is pulledup through the interface and into the oil phase. A similar equation can be used to determinethe interfacial tension from the force exerted on the ring, but one has to take into account the


FIGURE 5.14 Wilhelmy plate method of determining the interfacial and surface tension of liquids.

densities of the oil and water phases and use a different correction factor. For two liquidswhich are partially immiscible with each other, the interfacial tension may take an appre-ciable time to reach equilibrium because of the diffusion of water molecules into the oil phaseand vice versa (Hunter 1993).

The Du Nouy ring method can also be used to determine surface or interfacial tensions bycontinuously monitoring the force acting on the ring as the vessel containing the liquid islowered, rather than just measuring the detachment force. As the liquid is lowered, the forceinitially increases, but at a certain position it reaches a maximum (when the surface tensionacts vertically), before decreasing slightly prior to detachment. In this case, the maximumforce, rather than the detachment force, is used in the equations to calculate the surface orinterfacial tension (Couper 1993). The advantage of this method is that it does not involvethe rupture of the liquid, and therefore there are fewer problems associated with the kineticsof emulsifier adsorption during the detachment process.

For accurate measurements, it is important that the bottom edge of the ring be keptparallel to the surface of the fluid and that the contact angle between the liquid and the ringis close to zero. Rings are usually manufactured from platinum or platinum–iridium becausethese give contact angles that are approximately equal to zero. The Du Nouy ring methodcan be used to determine surface tensions to an accuracy of about 0.1 mN m–1 (Couper1993).

5.10.2. Wilhelmy Plate Method

The Wilhelmy plate method is normally used to determine the static surface or interfacialtensions of liquids (Hiemenz 1986, Couper 1993). Nevertheless, it can also be used tomonitor adsorption kinetics provided that the accumulation of the emulsifier at the surface isslow compared to the measurement time (Dickinson 1992). The apparatus consists of a vesselthat contains the liquid being analyzed and a plate which is attached to a sensitive force-measuring device (Figure 5.14). The vessel is capable of being moved upward and down-ward, while the plate remains stationary. The vessel is positioned so that the liquid just comesinto contact with the plate (i.e., the bottom of the plate is parallel to the surface of the bulkliquid). Some of the liquid “climbs” up the edges of the plate because this reduces theunfavorable contact area between the plate and the air. The amount of liquid which movesup the plate depends on its surface tension and density. If the force-measuring device is tarred


FIGURE 5.15 Sessile-drop and pendant-drop methods of determining the interfacial and surfacetension of liquids.

prior to bringing the plate into contact with the liquid (to take into account the mass of theplate), then the force recorded by the device is equal to the weight of the liquid clinging tothe plate. This weight is balanced by the vertical component of the surface tension multipliedby the length of the plate perimeter:

F l L= +2( ) cosγ θ (5.14)

where l and L are the length and thickness of the plate and θ is the contact angle. Thus thesurface tension of a liquid can be determined by measuring the force exerted on the plate.Plates are often constructed of materials which give a contact angle that is close to zero, suchas platinum or platinum/iridium, as this facilitates the analysis. The Wilhelmy plate can alsobe used to determine the interfacial tension between two liquids (Hiemenz 1986). In thiscase, the plate is positioned so that its bottom edge is parallel to the interface between thetwo bulk liquids, and the force-measuring device is tarred when the plate is located in theless dense liquid (usually oil). A major advantage of the Wilhelmy plate method over theDu Nouy ring method is that it does not rely on the disruption of the liquid surface andtherefore is less prone to errors associated with the adsorption kinetics of emulsifiers(Section 5.10.1).

The Wilhelmy plate method is widely used to study the adsorption kinetics of proteins(Dickinson 1992). It can be used for this purpose because their adsorption rate is much slowerthan the time required to carry out a surface tension measurement. The plate is positioned atthe surface of the liquid at the beginning of the experiment, and then the force required tokeep it at this position is recorded as a function of time. The Wilhelmy plate method is notsuitable for studying the adsorption kinetics of small-molecule surfactants because theyadsorb too rapidly to be followed using this technique. For accurate measurements, it isimportant that the bottom of the plate be kept parallel to the liquid surface and that the contactangle be either known or close to zero. Accuracies of about 0.05 mN m–1 have been reportedusing this technique (Couper 1993).

5.10.3. Sessile- and Pendant-Drop Methods

The sessile- and pendant-drop methods can be used to determine the static surface andinterfacial tensions of liquids (Hunter 1986, Couper 1993). The shape of a liquid dropletdepends on a balance between the gravitational and surface forces. Surface forces favor aspherical droplet because this shape minimizes the contact area between the liquid and itssurroundings. On the other hand, gravitational forces tend to cause droplets to becomeflattened (if they are resting on a solid surface) or elongated (if they are hanging on a solidsurface). A flattened drop is usually referred to as a sessile drop, whereas a hanging one isreferred to as a pendent drop (Figure 5.15). The equilibrium shape that is adopted by a dropletis determined by its volume, density, and surface or interfacial tension.


The surface or interfacial tension is determined by measuring the shape of a drop using anoptical microscope, often in conjunction with an image analysis program, and mathematicalequations which describe the equilibrium shape (Couper 1993). This technique can be usedto determine surface or interfacial tensions as low as 10–4 mN m–1 and can also be used tosimultaneously determine the contact angle. The accuracy of the technique has been reportedto be about 0.01% (Couper 1993).

5.10.4. Drop-Volume Method

The drop-volume method is used to measure surface and interfacial tensions of liquids by adetachment method. The liquid to be analyzed is pumped through the tip of a verticalcapillary tube whose tip protrudes into air or an immiscible liquid (Figure 5.16). A dropletdetaches itself from the tip of the capillary tube when the separation force (due to gravity)is balanced by the adhesion force (due to surface or interfacial tension). To measure surfacetension, the tip should point downward into air, whereas to measure interfacial tension, thetip may point either upward or downward depending on the relative densities of the twoliquids being analyzed. If the liquid in the tip has a higher density than the surroundingmedium, the opening of the tip faces down and the drop moves downward once it becomesdetached. On the other hand, if the liquid in the tip has a lower density, then the tip faces upand the drop moves upward when it becomes detached.

The surface or interfacial tension can be related to the volume of the detached droplet byanalyzing the forces which act on it just prior to detachment (Couper 1993):

γρ

π=

V g

dDROP∆

(5.15)

where d is the diameter of the tip, ∆ρ is the density difference between the liquid beinganalyzed and the surrounding liquid, VDROP is the volume of the droplet, and g is thegravitational constant. The volume of the droplets can be determined using a graduatedsyringe or by weighing them using a sensitive balance (once the liquid density is known).

As with other detachment methods, the surface area of the droplet expands rapidly duringthe detachment process, and therefore the method is unsuitable for analyzing liquids thatcontain emulsifiers whose adsorption time is comparable to the detachment time. To obtainreliable results, it is normal practice to bring the droplet slowly to the detachment process.The accuracy of this technique has been reported to be about 0.1% (Couper 1993).

FIGURE 5.16 Drop-volume method of determining the interfacial and surface tension of liquids.


5.10.5. Spinning-Drop Method

The spinning-drop method is used to measure static interfacial tensions of liquids (Couper1993). It was designed to characterize liquids that have particularly low interfacial tensions(i.e., between about 10–6 to 5 mN m–1). Most food emulsions have interfacial tensions thatare above this range, and so this technique is not widely used in the food industry. Never-theless, it may be useful for some fundamental studies, and therefore its operating principlesare briefly outlined below.

A drop of oil is injected into a glass capillary tube which is filled with a more dense liquid(usually water). When the tube is made to spin at a particular angular frequency (ω), thedroplet becomes elongated along its axis of rotation so as to reduce its rotational kineticenergy (Figure 5.17). The elongation of the droplet causes an increase in its surface area andis therefore opposed by the interfacial tension. Consequently, the equilibrium shape of thedroplet at a particular angular frequency depends on the interfacial tension. The shape of theelongated droplet is measured using optical microscopy, and the interfacial tension is thencalculated using the following equation:

γ ω ρ= kr 3 2∆ (5.16)

where k is an instrument constant which can be determined using liquids of known interfa-cial tension, r is the radius of the elongated droplet, and ∆ρ is the difference in densitybetween the oil and the surrounding liquid. This equation is applicable when the dropletbecomes so elongated that it adopts an almost cylindrical shape. More sophisticated equa-tions are needed when the shape of the droplet does not change as dramatically (Couper1993).

5.10.6. Maximum Bubble Pressure Method

The maximum bubble pressure method is used to measure the static or dynamic surfacetensions of liquids (Couper 1993). The apparatus required to carry out this type of measure-ment consists of a vertical capillary tube whose tip is immersed below the surface of theliquid being analyzed (Figure 5.18). Gas is pumped into the tube, which causes an increasein the pressure and results in the formation of a bubble at the end of the tip. The buildup ofpressure in the tube is monitored using a pressure sensor inside the instrument. As the bubblegrows, the pressure increases until it reaches a maximum when the bubble becomes hemi-spherical and the surface tension acts in a completely vertical direction. Any further bubble

FIGURE 5.17 Spinning-drop method of determining the interfacial tension of liquids.


FIGURE 5.18 Maximum bubble pressure method of determining the surface tension of liquids.

growth beyond this point causes a decrease in the pressure. Eventually, the bubble formedbreaks away from the tip and moves to the surface of the liquid; then another bubble beginsto form and the whole process is repeated. The maximum bubble pressure can be related tothe surface tension using suitable mathematical equations (Couper 1993).

The maximum bubble pressure method can be used to measure the static surface tensionof pure liquids at any bubble frequency. It can also be used to monitor the dynamic surfacetension of emulsifier solutions by varying the bubble frequency. The age of the gas–liquidinterface is approximately half the time interval between the detachment of successivebubbles and can be varied by changing the flow rate of the gas. It is therefore possible tomonitor adsorption kinetics of emulsifiers by monitoring the surface tension as a function ofbubble frequency. The dynamic surface tension increases as the bubble frequency increasesbecause there is less time for the emulsifier molecules to move to the surface of the droplets.The variation in the dynamic surface tension with bubble frequency therefore gives anindication of the speed at which emulsifier molecules are adsorbed to the surface. Thisinformation is important to food manufacturers because the adsorption rate of emulsifiersdetermines the size of the droplets produced during homogenization: the faster the rate, thesmaller the droplet size (Chapter 6). Nevertheless, it should be pointed out that the maximumbubble pressure method can only be used to measure dynamic surface tensions down to about50 ms, whereas many surfactants have faster adsorption rates than this. More rapid techniquesare therefore needed to study these systems, such as the oscillating jet method (Section5.10.7), the capillary wave method (Section 5.10.8), or the recently developed puncturedmembrane method (Stang et al. 1994).

One of the major advantages of the maximum bubble method is that it can be used toanalyze optically opaque liquids because it is not necessary to visually observe the bubbles.In addition, it is not necessary to know the contact angle of the liquid because the maximumpressure occurs when the surface tension acts in a completely vertical direction.

5.10.7. Oscillating Jet Method

The oscillating jet method is used to determine the static and dynamic surface and interfacialtensions of liquids (Couper 1993). The apparatus used for this type of measurement consistsof an elliptical orifice through which the liquid to be analyzed is pumped. The stream ofliquid that emerges changes its shape from a vertical elliptical cross-section to a sphericalcross-section as it moves away from the orifice because this enables it to reduce its surfacearea. Nevertheless, the liquid will continue to change its shape because of its momentum


FIGURE 5.19 Oscillating jet method of determining the interfacial and surface tension of liquids.

until it gains a horizontal elliptical cross-section (Figure 5.19). As the liquid travels awayfrom the orifice, its shape therefore oscillates between these different cross-sections. Theseoscillations have a characteristic wavelength which depends on the surface or interfacialtension of the liquid. The wavelength of the oscillating jet is measured using special opticaltechniques. The surface or interfacial tension can then be determined from the followingequation:

γπ ρ

λ=

2

3

2 3 2

2

r v

g (5.17)

where r is the radius of the orifice, v is the velocity of the liquid stream, g is the gravitationalconstant, and λ is the wavelength of oscillation. Surface tensions are measured by having thestream of liquid travel through air, whereas interfacial tensions are measured by having ittravel through an immiscible liquid. The age of the surface can be varied by altering the flowrate of the liquid emerging from the elliptical orifice: the faster the flow rate, the shorter theage of the interface. Dynamic surface or interfacial tensions can be measured down to timesof 1 ms using this method, and so it is particularly suitable for characterizing the adsorptionkinetics of rapidly adsorbing emulsifiers. Nevertheless, there is some debate about whetherthe movement of the emulsifier molecules to the surface is due to convection or diffusionunder the conditions prevailing within the instrument (Couper 1993). Pure liquids can beanalyzed at any flow rate, because their surface or interfacial tensions should be independentof surface age. The major limitation of this technique is that it can only be used to analyzelow-viscosity liquids (i.e., those with a viscosity less than about 5 mPa s).

5.10.8. Capillary Wave Method

The capillary wave method is a fairly new technique which can be used to measure the staticand dynamic surface tension of liquids (Couper 1993). A liquid is placed in a container thathas a device which is capable of generating surface waves at a certain frequency. Theamplitude and wavelength of the waves formed on the liquid surface are measured using alight-scattering technique and can be related to the surface tension and interfacial rheologyof the liquid using an appropriate theory. The surface age can be varied by altering thefrequency of the surface waves, and so it is possible to study adsorption kinetics. For pureliquids, the surface tension measured by this technique is independent of surface age and isin good agreement with that determined by static methods. For emulsifier solutions, thedynamic surface tension increases as the frequency of the surface waves is increased becausethe emulsifier molecules have less time to adsorb. The technique is particularly suitable foranalyzing liquids that have extremely low surface tensions and can be used to study surfaceages between about 5 and 100 ms.


TABLE 5.2Rheological Characteristics of Interfaces

Viscosity Elasticity

Shear deformation τ η σi i i= « τ σi i iG=

Dilatational deformation d Ad

dAi iγ κ γ= d

dAA

i iγ ε=

5.11. INTERFACIAL RHEOLOGY

Interfacial rheology has been defined as the “study of the mechanical and flow properties ofadsorbed layers at fluid interfaces” (Murray and Dickinson 1996). A knowledge of the factorswhich determine interfacial rheology is important to food scientists because the mechanicaland flow properties of interfaces influence the formation, texture, and stability of foodemulsions (Chapters 6 to 8). When an emulsion is subjected to mechanical agitation, thesurfaces of the droplets experience a number of different types of deformation as a result ofthe stresses acting upon them. Stresses may cause different regions of the interface to movepast each other, without altering the overall surface area, which is known as interfacial sheardeformation. Alternatively, they may cause the surface area to expand or contract (like aballoon when it is inflated or deflated), which is known as interfacial dilatational deforma-tion. An interface may have solid-like characteristics which are described by an interfacialelastic constant or fluid-like characteristics which are described by an interfacial viscosity.In practice, most interfaces have partly “solid-like” and partly “fluid-like” characteristics andtherefore exhibit viscoelastic behavior.

It is convenient to assume that an interface is infinitesimally thin so that it can be treatedas a two-dimensional plane, because its rheological characteristics can then be describedusing the two-dimensional analogs of the relationships used to characterize bulk materials(Chapter 8). A number of the most important rheological characteristics of interfaces aresummarized in Table 5.2, where τ i is the interfacial shear stress, σi is the interfacial shearstrain, ηi is the interfacial shear viscosity, Gi is the interfacial shear modulus, εi is theinterfacial dilatational elasticity, κ i is the interfacial dilatational viscosity, A is the surfacearea, and γ is the surface or interfacial tension.

The stability of emulsion droplets to coalescence is largely governed by the resistance ofthe oil–water interface to deformation and rupture (Chapter 7). Emulsifiers that form highlyviscoelastic membranes which are resistant to rupture often provide the greatest stabilityagainst coalescence. It is therefore important to have analytical techniques which can be usedto characterize the interfacial rheology of different emulsifiers and to establish the factors thatinfluence their interfacial properties.

5.11.1. Measurement of Interfacial Shear Rheology

A variety of experimental methods have been developed to measure the shear rheology ofsurfaces and interfaces (Murray and Dickinson 1996). One of the most commonly usedmethods is analogous to the concentric cylinder technique used to measure the shear prop-erties of bulk materials (Chapter 8). The sample to be analyzed is placed in a thermostatedvessel, and a thin disk is placed in the plane of the interface that separates the two phases (e.g.,water-and-air or oil-and-water) (Figure 5.20). The vessel is then rotated and the torque on thedisk is measured. The sample can be analyzed in a number of different ways depending onwhether it is solid-like, liquid-like, or viscoelastic. For liquid-like interfaces, the torque on the


FIGURE 5.20 Experimental technique for measuring interfacial shear rheology.

disk is measured as the vessel is rotated continuously. For solid-like interfaces, the torque ismeasured after the vessel has been moved to a fixed angle. For viscoelastic interfaces, thetorque is measured continuously as the vessel is made to oscillate backward and forward ata certain frequency and angle.

The interfacial shear viscosity or elasticity of surfactant membranes is usually severalorders of magnitude less than that of biopolymer membranes because biopolymer moleculesoften become entangled or interact with each other through various covalent or physicalforces. The rheology of bulk emulsions depends on the concentration, size, and interactionsof the droplets. By analogy, the rheology of interfaces depends on the concentration, size,and interactions of the adsorbed emulsifier molecules. Interfacial shear rheology measure-ments are particularly useful for providing information about adsorption kinetics, competi-tive adsorption, and the structure and interactions of molecules at an interface, especiallywhen they are used in conjunction with experimental techniques that provide informationabout the concentration of the emulsifier molecules at the interface (e.g., interfacial tensionor radioactive labeling techniques). The concentration of emulsifier molecules at an inter-face often reaches a constant value after a particular time, while the shear modulus orviscosity continues to increase because of interactions between the adsorbed molecules(Dickinson 1992).

5.11.2. Measurement of Interfacial Dilatational Rheology

Most analytical techniques used to characterize interfacial dilatational rheology measure thechange in interfacial tension as the surface area is increased or decreased in a controlledmanner (e.g., trough and overflowing cylinder methods) (Murray and Dickinson 1996).Trough methods measure the surface or interfacial tension of a liquid using a Wilhelmyplate as the interfacial area is varied by changing the distance between two solid barrierswhich confine the liquid (Figure 5.21). The overflowing cylinder method can be used tomeasure the dynamic dilatational rheology of surfaces or interfaces. A Wilhelmy plate isused to measure the surface or interfacial tension of a liquid as it is continuously pumpedinto a cylinder so that it overflows at the edges. The dilatational viscosity is measured as afunction of the surface age by altering the rate at which the liquid is pumped into thecylinder.

The capillary wave method described in Section 5.10.8 can also be used to measure theinterfacial dilatational rheology of liquids. A laser beam reflected from the surface of a liquidis used to determine the amplitude and wavelength of the surface waves, which can then berelated to the dilatational modulus or viscosity of the surface using an appropriate theory.These surface waves are believed to play an important role in the coalescence of emulsiondroplets (Chapter 7), and therefore this technique may provide information that has directpractical importance for food scientists.


FIGURE 5.21 Experimental technique for measuring interfacial dilatational rheology.

It should be noted that the dilatational rheology of an interface can be influenced by theadsorption kinetics of emulsifiers (Murray and Dickinson 1996). When a surface undergoesa dilatational expansion, the concentration of emulsifier per unit area decreases and thereforeits surface tension increases, which is energetically unfavorable. The dilatational elasticity orviscosity is a measure of this resistance of the surface to dilation (Table 5.2). When emulsifiermolecules are present in the surrounding liquid, they may be adsorbed to the surface duringdilation and thereby reduce the surface tension. The dilatational rheology therefore dependson the rate at which emulsifier molecules are adsorbed to a surface, which is determined bytheir concentration, molecular structure, and the prevailing environmental conditions (Section5.4). The faster the molecules adsorb to the freshly formed interface, the lower is theresistance to dilation, and therefore the lower is the dilatational modulus or viscosity. Whenan interface undergoes dilatational compression, some of the emulsifier molecules may leavethe interface to reduce the resulting strain, and therefore the desorption rate may also influ-ence the rheological characteristics of an interface.

5.12. CHARACTERIZATION OF INTERFACIAL STRUCTURE

The structural organization of the emulsifier molecules at the surface of the droplets in anemulsion influences the properties of the whole emulsion. Consequently, it is important forfood scientists to have analytical techniques that can be used to provide information aboutinterfacial structure, such as the thickness of the adsorbed layer and the packing andorientation of the emulsifier molecules (Dalgleish 1996). A number of the most importantexperimental techniques that are available to provide information about interfacial struc-ture are listed in Table 5.3. Each of these techniques works on a different physicalprinciple and is sensitive to a different aspect of the properties of an interface (Kallay et al.1993).

A variety of experimental techniques rely on the scattering of radiation or subatomicparticles from a surface or interface (e.g., light, X-ray, or neutron scattering techniques).Some of these techniques can only be applied to planar interfaces, while others can also beapplied to the study of interfaces on emulsion droplets. Analytical techniques based on thescattering of radiation or subatomic particles can be used to provide information about thethickness of the interfacial layer and sometimes also about the variation in the emulsifierconcentration within the interface. They can therefore be used to compare the thickness ofinterfaces formed by different emulsifiers or to study changes in interfacial thickness inresponse to alterations in environmental conditions such as pH, ionic strength, or temperature.The application of these techniques has provided valuable insights into the structures formed


TABLE 5.3Experimental Techniques to Characterize the Structural Propertiesof Interfaces

Techniques Applications

Scattering techniquesLight scattering ThicknessSmall-angle X-ray scattering Thickness, concentration profileNeutron scattering Thickness, concentration profile

Reflection techniquesEllipsometry Thickness, refractive indexNeutron reflection Thickness, concentration profile

Techniques that utilize absorptionof electromagnetic radiation

Infrared Concentration, conformational changesUltraviolet-visible Concentration, conformational changesFluorescence Concentration, conformational changesNuclear magnetic resonance Concentration, conformational changesCircular dichroism Secondary structure

Biochemical techniquesEnzyme hydrolysis Orientation, conformationImmunoassays Orientation, conformation

Microscopic techniquesAtomic force microscopy Structural organization of moleculesElectron microscopy Interfacial structure

by proteins at oil–water and air–water interfaces, which is useful for predicting the stabilityof emulsion droplets to flocculation or coalescence (Dalgleish 1996).

A number of spectroscopic techniques have been developed that rely on the absorption ofelectromagnetic radiation by the molecules at an interface (e.g., infrared, ultraviolet, visible,fluorescence, and X-ray techniques) (Kallay et al. 1993). These techniques can be used eitherin a transmission mode, where the electromagnetic wave is propagated through the sample,or in a reflection mode, where the electromagnetic wave is reflected from the interface. Inboth cases, the absorption of the radiation by the wave is measured and correlated to someproperty of the molecules at the interface. The different techniques are sensitive to differentcharacteristics of the molecules. Infrared relies on characteristic molecular vibrations ofdifferent groups within a molecule, whereas ultraviolet-visible and fluorescence techniquesrely on characteristic electronic transitions which occur in unsaturated groups. These tech-niques have been used to determine the concentration of emulsifier molecules at an interface,to study adsorption kinetics, and to provide information about changes in molecular confor-mation (Kallay et al. 1993). When a molecule undergoes a conformational change, theadsorption spectrum is altered because of the change in the environment of the chemicalgroups.

A number of techniques have been developed which rely on the interference patternproduced by the reflection of a beam of light or subatomic particles from a thin interface.Ellipsometry relies on the reflection of a polarized light beam from a highly reflective planarsurface (Azzam and Bachara 1989, Malmsten et al. 1995). Initially, the phase and amplitudeof the light beam reflected from the clean surface are measured. Emulsifier is then allowedto adsorb to the surface and the experiment is repeated. A mathematical analysis of thereflected wave provides information about the thickness and refractive index profile of theadsorbed layer. The refractive index profile can then be converted to a concentration profile.


Neutron reflection techniques work on a similar principle, but they use a beam of neutrons,rather than a beam of light, to probe the interface (Atkinson 1995, Dickinson and McClements1995). The beam of neutrons is directed at an air–water or oil–water interface at an angle. Thebeam of neutrons reflected from the surface is analyzed and provides information about thethickness of the interface and the neutron refractive index profile, which can be related to theemulsifier concentration profile across the interface. These techniques have been widely usedto study the characteristics of protein layers and the influence of pH and ionic strength ontheir properties.

A number of biochemical techniques have also been developed to provide informationabout the structure and organization of biopolymer molecules at the surface of emulsiondroplets (Dalgleish 1996b). The susceptibility of proteins to hydrolysis by specific enzymescan be used to identify the location of particular peptide bonds. When a protein is dissolvedin an aqueous solution, the whole of its surface is accessible to enzyme hydrolysis, but whenit is adsorbed to an interface, it adopts a conformation where some of the amino acids arelocated close to the interface (and are therefore inaccessible to proteolysis), whereas othersare exposed to the aqueous phase (and are therefore accessible to proteolysis). By comparingthe bonds which are susceptible to enzyme hydrolysis and the rate at which hydrolysis of theadsorbed and unadsorbed proteins occurs, it is possible to obtain some idea about theorientation and conformation of the adsorbed proteins at the interface. This technique is moresuitable for the study of flexible random-coil proteins than globular proteins, because thelatter have compact structures which are not particularly accessible to enzyme hydrolysis ineither the adsorbed or unadsorbed state. Immunological techniques can also be used toprovide similar information. These techniques utilize the binding of antibodies to specificsites on a protein to provide information about its orientation at an interface or about anyconformational changes which take place after adsorption.

The thickness of a layer of emulsifier molecules adsorbed to an interface can be determinedby measuring the forces between two interfaces as they are brought into close contact(Claesson et al. 1995, 1996). There is usually a steep rise in the repulsive forces at closeseparations between the interfaces because of steric repulsion. The distance at which this riseis observed can be assumed to be the outer edge of the emulsifier layer.

Information about the conformation of proteins adsorbed to the surfaces of emulsiondroplets has also been obtained using sensitive differential scanning calorimetry (DSC)techniques (Corredig and Dalgleish 1995; Dalgleish 1996a,b). The emulsion being analyzedis heated at a controlled rate and the amount of heat absorbed by the proteins when theyundergo conformational changes is recorded by the DSC instrument. The temperature atwhich the conformation change occurs (Ttransition) and the amount of energy adsorbed by theprotein during the transition (∆Htransition) are determined. The values for the adsorbed proteinare then compared with those for a solution containing a similar concentration of unadsorbedproteins. Experiments with oil-in-water emulsions containing adsorbed globular proteins,such as α-lactalbumin, lysozyme, and β-lactoglobulin, have shown that there is a decrease in∆Htransition after adsorption, which indicates that they have undergone some degree of surfacedenaturation (Corredig and Dalgleish 1995). It was also observed that after the proteins hadbeen desorbed from the surface, the denaturation of β-lactoglobulin was irreversible, but thatof α-lactalbumin and lysozyme was at least partly reversible. Flexible biopolymers that havelittle structure, such as casein, cannot be studied using this technique because they do notundergo any structural transitions which adsorb or release significant amounts of heat. Itshould be noted that the data from DSC experiments must be interpreted carefully becausea reduction in ∆Htransition may be because the protein has no secondary or tertiary structure orbecause the structure is so stable that it does not unfold at the same temperature as in water.In addition, the adsorption of a protein at an interface may also reduce ∆Htransition because its


thermodynamic environment has been altered: in solution, it is completely surrounded bywater, but at an interface, part of the molecule is in contact with a hydrophobic surface.Despite these limitations, the DSC technique is a powerful tool for studying structuralchanges caused by adsorption, especially when used in conjunction with other techniques.

5.13. PRACTICAL IMPLICATIONS OF INTERFACIAL PHENOMENA

To conclude this chapter, a brief outline of some of the most important practical implicationsof interfacial phenomena for food emulsions is presented. One of the most striking featuresof food emulsions when observed under a microscope is the sphericity of the droplets.Droplets tend to be spherical because this shape minimizes the energetically unfavorablecontact area between oil and water molecules, which is described by the Laplace equation(Equation 5.4). Droplets become nonspherical when they experience an external force that islarge enough to overcome the Laplace pressure (e.g., gravity, centrifugal forces or mechanicalagitation). The magnitude of the force needed to deform a droplet decreases as the interfacialtension decreases or the radius increases. This accounts for the ease with which the relativelylarge droplets in highly concentrated emulsions, such as mayonnaise, are deformed intopolygons (Dickinson and Stainsby 1982).

The thermodynamic driving force for coalescence and “oiling off” is the interfacial tensionbetween the oil and water phases caused by the imbalance of molecular interactions acrossan oil–water interface (Section 7.2). The reason that surface-active molecules adsorb to anair–fluid or oil–water interface is because of their ability to reduce the surface or interfacialtension (Section 5.2.2.1). The tendency of a liquid to spread over the surface of anothermaterial or to remain as a lens depends on the relative magnitude of the interfacial and surfacetensions between the different types of substance involved (Section 5.5).

The formation of stable nuclei in a liquid is governed by the interfacial tension betweenthe crystal and the melt (Section 4.2). The larger the interfacial tension between the solid andliquid phases, the greater the degree of supercooling required to produce stable nuclei. Theinterfacial tension also determines whether an impurity is capable of promoting heterogenousnucleation.

The solubility of a substance increases as the size of the particle containing it decreases(Section 5.4). If a suspension contains particles (emulsion droplets, fat crystals, ice crystals,or air bubbles) of different sizes, there is a greater concentration of the substance dissolvedin the region immediately surrounding the smaller particles than that surrounding the largerparticles. Consequently, there is a concentration gradient which causes material to move fromthe smaller particles to the large particles. With time, this process manifests itself as a growthof the large particles at the expense of the smaller ones, which is referred to as Ostwaldripening (Chapter 7). This process is responsible for the growth in size of emulsion droplets,fat crystals, ice crystals, and air bubbles in food emulsions, which often has a detrimentaleffect on the quality of a food. For example, the growth of ice crystals in ice cream causesthe product to be perceived as “gritty” (Berger 1976). It should be clear from this chapter thateven though the interfacial region only comprises a small fraction of the total volume of anemulsion, it plays an extremely important role in determining its bulk physicochemicalproperties.


6 Emulsion Formation

6.1. INTRODUCTION

Fresh milk is an example of a naturally occurring emulsion that can be consumed directly byhuman beings (Swaisgood 1996). In practice, however, most milk is subjected to a numberof processing operations prior to consumption in order to ensure its safety, to extend its shelflife, and to create new products. Processing operations, such as homogenization, pasteuriza-tion, whipping, and churning, are responsible for the wide range of properties exhibited bydairy products (e.g., homogenized milk, cream, ice cream, butter, and cheese) (Harper andHall 1976, Desrosier 1977, Kessler 1981). Unlike dairy products, most other food emulsionsare manufactured by combining raw materials which are not normally found together innature (Friberg and Larsson 1997, Dickinson and Stainsby 1982). For example, a saladdressing may be prepared using water, proteins from milk, oil from soybeans, vinegar fromapples, and polysaccharides from seaweed. The physicochemical and sensory properties of aparticular food emulsion depend on the type and concentration of ingredients it contains, aswell as the method used to create it. To improve the quality of existing products, develop newproducts, and reduce production costs, it is important for food manufacturers to have athorough understanding of the physical processes which take place during emulsion forma-tion. The physical principles of emulsion formation, the various techniques available forcreating emulsions, and the factors which affect the efficiency of emulsion formation arediscussed in this chapter.

6.2. OVERVIEW OF HOMOGENIZATION

The formation of an emulsion may involve a single step or a number of consecutive steps,depending on the nature of the starting material and the method used to create it. Prior toconverting separate oil and aqueous phases into an emulsion, it is usually necessary todisperse the various ingredients into the phase in which they are most soluble. Oil-solubleingredients, such as vitamins, colors, antioxidants, and surfactants, are usually mixed with theoil, whereas water-soluble ingredients, such as proteins, polysaccharides, sugars, salts, vita-mins, colors, antioxidants, and surfactants, are usually mixed with the water. The intensityand duration of the mixing process depend on the time required to solvate and uniformlydistribute the ingredients. Adequate solvation is important for the functionality of a numberof food components (e.g., the emulsifying properties of proteins are often improved byallowing them to hydrate in water for a few minutes or hours prior to homogenization)(Kinsella and Whitehead 1989). If the lipid phase contains any crystalline material, it isnecessary to warm it to a temperature where all the fat melts prior to homogenization;otherwise it is difficult, if not impossible, to create a stable emulsion (Mulder and Walstra1974, Phipps 1985).

The process of converting two immiscible liquids into an emulsion is known as homogeni-zation, and a mechanical device designed to carry out this process is called a homogenizer


* In the presence of an emulsifier, the type of emulsion formed is governed mainly by the properties of theemulsifier (i.e., the HLB number and optimum curvature) (Chapter 4).

FIGURE 6.1 Homogenization can be conveniently divided into two categories: primary and second-ary. Primary homogenization is the conversion of two bulk liquids into an emulsion, whereas secondaryhomogenization is the reduction in size of the droplets in an existing emulsion.

(Loncin and Merson 1979). To distinguish between the nature of the starting materials, it isconvenient to divide homogenization into two categories. The creation of an emulsion di-rectly from two separate liquids will be defined as primary homogenization, whereas thereduction in size of the droplets in an existing emulsion will be defined as secondaryhomogenization (Figure 6.1). The creation of a particular type of food emulsion may involvethe use of either of these types of homogenization or a combination of both. For example, thepreparation of a salad dressing in the kitchen is carried out by direct homogenization of theaqueous and oil phases and is therefore an example of primary homogenization, whereashomogenized milk is manufactured by reducing the size of the fat globules in raw milk andso is an example of secondary homogenization. In many food-processing operations andlaboratory studies, it is more efficient to prepare an emulsion using two steps (Dickinson andStainsby 1982). The separate oil and water phases are converted to a coarse emulsion whichcontains fairly large droplets using one type of homogenizer (e.g., a high-speed blender), andthen the size of the droplets is reduced using another type of homogenizer (e.g., a high-pressure valve homogenizer). Many of the same physical processes occur during primary andsecondary homogenization (e.g., mixing, droplet disruption, and droplet coalescence), and sothere is no clear distinction between the two.

Emulsions which have undergone secondary homogenization usually contain smaller drop-lets than those which have undergone primary homogenization, although this is not alwaysthe case. Some homogenizers are capable of producing emulsions with small droplet sizesdirectly from the separate oil and water phases (e.g., ultrasound, microfluidizers, or mem-brane homogenizers) (see Section 6.4).

The physical processes which occur during homogenization can be highlighted by consid-ering the formation of an emulsion from pure oil and pure water. When the two liquids areplaced in a container, they tend to adopt their thermodynamically most stable state, whichconsists of a layer of oil on top of a layer of water (Figure 6.1). This arrangement is adoptedbecause it minimizes the contact area between the two immiscible liquids and because oil hasa lower density than water (Section 7.2). To create an emulsion, it is necessary to supplyenergy in order to disrupt and intermingle the oil and water phases, which is usually achievedby mechanical agitation (Walstra 1993b). The type of emulsion formed in the absence of anemulsifier depends primarily on the initial concentration of the two liquids: at high oilconcentrations, a water-in-oil emulsion tends to be formed, but at low oil concentrations, anoil-in-water emulsion tends to be formed.* In this example, we assume that the oil concen-tration is so low that an oil-in-water emulsion is formed. Mechanical agitation can be applied


in a variety of different ways (Section 6.4), the simplest being to vigorously shake the oil andwater together in a sealed container. Immediately after shaking, an emulsion is formed whichappears optically opaque because light is scattered by the emulsion droplets (Farinato andRowell 1983). The oil droplets formed during the application of the mechanical agitation areconstantly moving around and frequently collide and coalesce with neighboring droplets(Walstra 1993b). As this process continues, the large droplets formed move to the top of thecontainer due to gravity and merge together to form a separate layer. As a consequence, thesystem reverts back to its initial state — a layer of oil sitting on top of a layer of water (Figure6.1). The thermodynamic driving forces for this process are the hydrophobic effect, whichfavors minimization of the contact area between the oil and water, and gravity, which favorsthe upward movement of the oil (Section 7.2).

To form an emulsion that is (kinetically) stable for a reasonable period of time, one mustprevent the droplets from merging together after they have been formed (Walstra 1983,1993b). This is achieved by having a sufficiently high concentration of emulsifier presentduring the homogenization process. The emulsifier adsorbs to the surface of the dropletsduring homogenization, forming a protective membrane which prevents them from comingclose enough together to coalesce. The size of the droplets produced during homogenizationdepends on two processes: (1) the initial generation of droplets of small size and (2) the rapidstabilization of these droplets against coalescence once they are formed (Section 6.3).

Many of the bulk physicochemical and organoleptic properties of food emulsions dependon the size of the droplets they contain, including their stability, texture, appearance, and taste(Chapters 7 to 9). One of the major objectives of homogenization is therefore to create anemulsion in which the majority of droplets fall within an optimum range which has previouslybeen established by the food manufacturer. It is therefore important for food scientists toappreciate the major factors which determine the size of the droplets produced duringhomogenization.

This brief introduction to homogenization has highlighted some of the most importantaspects of emulsion formation, including the necessity to mechanically agitate the system, thecompeting processes of droplet formation and droplet coalescence, and the role of theemulsifier. These topics will be considered in more detail in the rest of the chapter.

6.3. PHYSICAL PRINCIPLES OF EMULSION FORMATION

As mentioned in the previous section, the size of the droplets produced by a homogenizerdepends on a balance between two opposing physical processes: droplet disruption anddroplet coalescence (Figure 6.2). A better understanding of the factors which influence theseprocesses would enable food manufacturers to select the most appropriate ingredients andhomogenization conditions required to produce a particular food product.

6.3.1. Droplet Disruption

The precise nature of the physical processes which occur during emulsion formation dependson the type of homogenizer used. Nevertheless, there are some common aspects of dropletdisruption which apply to most types of homogenizer. The initial stages of primary homog-enization involve the breakup and intermingling of the bulk oil and aqueous phases so thatfairly large droplets of one of the liquids become dispersed throughout the other liquid(Walstra 1983, 1993b). The later stages of primary homogenization, and the whole of second-ary homogenization, involve the disruption of larger droplets into smaller ones. It is thereforeimportant to understand the nature of the forces which are responsible for the disruption ofdroplets during homogenization. Whether or not a droplet breaks up is determined by abalance between interfacial forces, which tend to hold the droplets together, and the disrup-


FIGURE 6.2 The size of the droplets produced during homogenization depends on the balancebetween the time for an emulsifier to adsorb to the surface of a droplet (τ adsorb) and the time betweendroplet–droplet collisions (τ collision).

tive forces generated within the homogenizer, which tend to pull them apart (Walstra 1983,1993b).

6.3.1.1. Interfacial Forces

An emulsion droplet tends to be spherical because this shape minimizes the energeticallyunfavorable contact area between the oil and aqueous phases (Chapter 5). Changing the shapeof a droplet, or breaking it up into a number of smaller droplets, increases this contact areaand therefore requires an input of energy. The interfacial force responsible for keeping adroplet in a spherical shape is characterized by the Laplace pressure (∆PL), which acts acrossthe oil–water interface toward the center of the droplet so that there is a larger pressure insidethe droplet than outside of it:

∆PdL =4γ

(6.1)

Here, γ is the interfacial tension between oil and water, and d is the droplet diameter. Todeform and disrupt a droplet during homogenization, it is necessary to apply an external forcewhich is significantly larger than the interfacial force (Walstra 1983, 1996b). Equation 6.1indicates that the pressure required to disrupt a droplet grows as the interfacial tensionincreases or as the droplet size decreases. It also indicates that intense pressures must begenerated within a homogenizer in order to overcome the interfacial forces holding theemulsion droplets together. For example, the Laplace pressure of a 1-µm droplet with aninterfacial tension of 0.01 N m–1 is about 40 kPa, which corresponds to a pressure gradientof ∆PL/d ≈ 40 × 109 Pa m–1 across the droplet.

6.3.1.2. Disruptive Forces

The nature of the disruptive forces which act on a droplet during homogenization depends onthe flow conditions it experiences (i.e., laminar, turbulent, or cavitational) and therefore onthe type of homogenizer used (Phipps 1985, Walstra 1993b). For a droplet to be broken upduring homogenization, the disruptive forces must exceed the interfacial forces, and theirduration must be longer than the time required to deform and disrupt the droplet (τdeform)


FIGURE 6.3 In the presence of a simple shear flow, droplets may rotate and become elongated. Inaddition, the fluid inside a droplet may circulate around the center of the droplet.

(Stone 1994, Karbstein and Schubert 1995). The susceptibility of emulsion droplets todisruption is conveniently characterized by the Weber number (We), which is the ratio of thedisruptive forces to the interfacial forces (Walstra 1983). Droplets are disrupted when theWeber number exceeds some critical value (around unity), which depends on the physicalcharacteristics of the oil and aqueous phases.

The flow profile of an emulsion within a homogenizer is usually extremely complex andis therefore difficult to model mathematically (Phipps 1985). For this reason, it is not easyto accurately calculate the disruptive forces that a droplet experiences during homogeniza-tion. Nevertheless, it is possible to gain much insight into the factors which affect the homog-enization process by considering droplet disruption under simpler flow conditions whichapproximate those that occur in actual homogenizers (i.e., laminar, turbulent, or cavitationalflow conditions) (Gopal 1968; Walstra 1983, 1993b; Williams et al. 1997).

Laminar Flow Conditions. This type of flow profile is predominant at low flow rates, wherethe fluid moves in a regular and well-defined pattern (Curle and Davies 1968, Walstra1993b). Different types of laminar flow profile are possible, depending on the direction andvelocity at which different regions within the fluid move relative to one another (e.g., parallel,simple shear, rotational, and hyperbolic flow) (Gopal 1968). For convenience, only thedisruption of droplets under simple shear flow conditions is considered here. In the presenceof a simple shear field, a droplet experiences a combination of normal and tangential stresses(Loncin and Merson 1979). These stresses cause the droplet to rotate and become elongated,as well as causing the liquid within the droplet to circulate (Figure 6.3). At sufficiently highshear rates, the droplet becomes so elongated that it is broken up into a number of smallerdroplets (Stone 1994, Williams et al. 1997). The manner in which the droplets break updepends on the ratio of the viscosities of the droplet and continuous phase (ηD/ηC). Experi-ments in which droplets were photographed under different flow conditions have shown thatat low values of ηD/ηC the droplets break up at their edges, at intermediate values they breakup near their middle, and at high values they may not break up at all, because there isinsufficient time for the droplets to deform during the application of the disruptive forces(Gopal 1968, Williams et al. 1997).

The disruptive forces that a droplet experiences during simple shear flow are determinedby the shear stress (GηC) which acts upon the droplet, and so the Weber number is given by(Walstra 1993b):*

* The reason that a factor of 4 appears in Equation 6.3, whereas a factor of 2 appears in Equation 6.2, is becauseonly half of the applied shear force goes to deforming the droplet; the remainder causes the droplet to rotate andis therefore not responsible for droplet disruption.


FIGURE 6.4 Dependence of the critical Weber number on the viscosity of the dispersed and continu-ous phases under simple shear flow conditions. (Adapted from Schubert and Armbruster 1992.)

Weshear forces

interfacial forces= =

G dCηγ2 (6.2)

Here, G is the shear rate and ηC is the viscosity of the continuous phase. For a givensystem, it is possible to define a critical Weber number (Wecritical), which is the value of Wewhere the droplets are just stable to disruption. If an emulsion has a Weber number above thiscritical value (i.e., high shear rates or large droplets), then the droplets will be broken up;otherwise they will remain intact (Figure 6.4). The critical Weber number of an emulsion inthe absence of emulsifier depends principally on the ratio of the viscosities of the dispersedand continuous phases (Janssen et al. 1994, Karbstein and Schubert 1995, Williams et al.1997). Wecritical has a minimum value when ηD/ηC is between about 0.1 and 1 and increasessignificantly as the viscosity ratio decreases below about 0.05 or increases above about 5(Figure 6.4). The behavior of droplets during the disruption process has been widely studiedand is now well understood (Bentley and Leal 1986, Stone 1994, Janssen et al. 1994).Droplets are resistant to breakup at low viscosity ratios (<0.05) because they are able tobecome extremely elongated before any disruption will occur. They are resistant to breakupat high viscosity ratios (>5) because they do not have sufficient time to become deformedbefore the flow field causes them to rotate to a new orientation and therefore alter thedistribution of disruptive stresses acting on them. At intermediate viscosity ratios, the drop-lets tend to form a dumbbell shape just prior to breaking up.

In the presence of emulsifiers, such as small-molecule surfactants or proteins, the behaviorof droplets in flow fields has been found to be different from that in the absence of emulsi-fiers, which has been attributed to their influence on the rheology of the interfacial membrane(Lucassen-Reyanders and Kuijpers 1992, Janssen et al. 1994, Williams et al. 1997). Dropletsare more difficult to disrupt than would be expected from their equilibrium interfacial tension,because the emulsifier imparts rheological properties to the droplet interface, which increasesits resistance to tangential stresses.

A food manufacturer usually wants to produce an emulsion which contains droplets belowsome specified size, and so it is important to establish the factors which determine the sizeof the droplets produced during homogenization. For simple shear flow, the following rela-tionship gives a good description of the maximum size of droplets which can persist in anemulsion during homogenization under steady-state conditions (Walstra 1983, 1993b):

dC

max =2γ

ηWe

Gcritical

(6.3)


Any droplets larger than dmax will be disrupted, whereas any smaller droplets will remainintact. Equation 6.3 indicates that the size of the droplets produced during homogenizationdecreases as the interfacial tension (γ) decreases, as the shear rate increases, or as the viscosityof the continuous phase increases. It also indicates that a higher shear rate is required todecrease the droplet size when the viscosity of the continuous phase is low. It is for thisreason that homogenizers which rely principally on simple shear flow conditions, such ascolloid mills, are not suitable for generating emulsions with small droplet sizes when thecontinuous phase has a low viscosity (Walstra 1983). For this type of system, it is better touse homogenizers which utilize turbulent or cavitational conditions to break up the droplets.

Turbulent Flow Conditions. Turbulence occurs when the flow rate of a fluid exceeds somecritical value, which is largely determined by its viscosity (Curle and Davies 1968; Walstra1983, 1993b; Phipps 1985). Turbulence is characterized by rapid and chaotic fluctuations inthe velocity of the fluid with time and location. The disruption of droplets under turbulentflow conditions is caused by the extremely large shear and pressure gradients associated withthe eddies generated in the fluid (Walstra 1993b). An eddy is a region within a fluid wherethere is a close correlation between the fluid velocity of the different elements (Curle andDavies 1968). Normally, a range of different-sized eddies is formed within a liquid duringturbulence. The shear and pressure gradients associated with these eddies increase as theirsize decreases (Walstra 1983). As a consequence, large eddies are believed to be relativelyineffective at disrupting emulsion droplets. Very small eddies are also believed to be ineffec-tive at breaking up droplets because they generate such high shear stresses that most of theirenergy is dissipated through viscous losses rather than through droplet disruption. For thesereasons, intermediate-sized eddies are thought to be largely responsible for droplet disruptionunder turbulent flow conditions (Walstra 1983, 1993b). When a droplet is in the vicinity ofone of these intermediate-sized eddies, it is deformed and disrupted because of the large sheargradient acting across it (Gopal 1968). For isotropic turbulent conditions, the Weber numberis given by (Karbstein and Schubert 1995):

Weturbulent forcesinterfacial forces

= =C dCρ ε

γ

1 3 2 3 5 3

4

/ / /

(6.4)

where C is a constant related to the critical Weber number, ρC is the density of the continuous

phase, and ε is the power density. Under isotropic turbulent conditions, the maximum size ofdroplets that can persist during homogenization once a steady state has been reached has beenshown to be given by the following relationship (Walstra 1983):

dC

Cmax

/

/ /=

′γε ρ

3 5

2 5 1 5 (6.5)

where C′ is a constant which depends on the characteristic dimensions of the system. Thisequation indicates that the size of the droplets produced under turbulent conditions decreasesas the power density increases, the interfacial tension decreases, or the density of the continu-ous phase increases. It also suggests that the viscosity of the liquids does not influence thedroplet size. Nevertheless, a number of experimental studies have shown that the viscositiesof the dispersed and continuous phases do influence the maximum droplet size that can persistduring homogenization, with a minimum in dmax when ηD/ηC is between about 0.1 and 5(Braginsky and Belevitskaya 1996). The dependence of the droplet size produced duringhomogenization on the viscosity ratio under turbulent flow conditions is therefore similar in


form to that produced under laminar flow conditions (Figure 6.4). It is therefore possible toreduce the size of the droplets produced during homogenization by ensuring that the viscosityratio falls within the optimum range for droplet breakup (0.1 < ηD/ηC < 5), which could beachieved by varying the temperature or by adding thickening agents. The viscosity may alsoinfluence the droplet size if it is large enough to suppress turbulence. For example, anincrease in viscosity due to the presence of thickening agents or high concentrations ofdroplets may be sufficient to prevent turbulent flow conditions and therefore lead to ineffi-cient homogenization (Walstra 1993b).

An emulsion does not normally remain in a homogenizer long enough for steady-stateconditions to be attained, and so Equation 6.5 is not strictly applicable. In practice, the sizeof the droplets in an emulsion decreases as the length of time they spend in the disruptionzone of a homogenizer increases, until eventually a constant value is reached (Karbstein andSchubert 1995). This is because droplets take a finite time to be deformed, and so theturbulent forces must act over a period which is sufficiently longer than this time if all thedroplets are to be effectively disrupted (Walstra 1993b). The deformation time is proportionalto the viscosity of a droplet; therefore, the more viscous the dispersed phase, the less likelyis droplet breakup within a specified time. Emulsions produced under turbulent flow condi-tions are always polydisperse because of the distribution of eddy sizes in the fluid. In fact,the statistical theories used to derive the above equations indicate that droplets formed underturbulent conditions should follow a log-normal distribution, which is often observed inpractice (Gopal 1968).

Cavitational Flow Conditions. Cavitation occurs in fluids which are subjected to rapidchanges in pressure and is particularly important in ultrasonic and high-pressure valvehomogenizers (Gopal 1968, Phipps 1985). A fluid contracts when the pressure acting on itincreases and expands when the pressure decreases. When the instantaneous pressure thata fluid experiences falls below some critical value, a cavity is formed (Lickiss and McGrath1996). As the fluid continues to expand, the cavity grows and some of the surrounding liquidevaporates and moves into it. During a subsequent compression, the cavity catastrophicallycollapses, generating an intense shock wave which propagates into the surrounding fluid andcauses any droplets in its immediate vicinity to be deformed and disrupted. Extremely hightemperatures and pressures are associated with these shock waves, but they are of very shortduration and highly localized, so that little damage is usually done to the vessel containingthe fluid. Nevertheless, over time, cavitational effects are known to cause significant damageto the surfaces of high-pressure valve homogenizers and ultrasonic transducers, whichbecome “pitted” (Gopal 1968, Phipps 1985). Cavitation only occurs in fluids when theintensity of the fluctuating pressure field exceeds a critical value, known as the cavitationalthreshold. This threshold is high in pure liquids, but is reduced when cavitational nuclei,such as gas bubbles or impurities, are present. The cavitational threshold also depends onthe frequency of the pressure fluctuations, decreasing with decreasing frequency (Gopal1968).

Homogenization of Nonideal Liquids. The equations given above are only strictly appli-cable to homogenization of ideal (Newtonian) liquids (Chapter 8). In practice, many liquidsused in the food industry exhibit nonideal behavior, which can have a pronounced influenceon the efficiency of droplet disruption and therefore on the size of the droplets producedduring homogenization (Walstra 1983, 1993b). Many biopolymers used to thicken or stabi-lize emulsions exhibit pronounced shear thinning behavior (Chapters 4 and 8). As a conse-quence, the viscosity used in the above equations should be that which the droplet experi-ences at the shear rates that occur during homogenization, rather than that which is measuredin a viscometer at low shear rates.


6.3.1.3. The Role of the Emulsifier in Droplet Disruption

The ease with which a droplet can be disrupted during homogenization increases as theinterfacial tension decreases (Equation 6.1). Thus it is possible to produce droplets withsmaller sizes by homogenizing in the presence of an emulsifier which reduces the interfacialtension (Walstra 1993b). For example, adding an emulsifier that decreases the interfacialtension from 50 to 5 mN m–1 should decrease the size of the droplets produced under laminarflow conditions by an order of magnitude (Schubert and Armbruster 1995). Nevertheless, anumber of other factors also determine the effectiveness of emulsifiers at reducing the dropletsize. First, the rate at which an emulsifier adsorbs to the surface of the droplets duringhomogenization must be considered (Walstra 1983). Immediately after their formation, drop-lets have a low concentration of emulsifier adsorbed to their surface and are therefore moredifficult to disrupt because of the relatively high interfacial tension. With time, a greateramount of emulsifier accumulates at the surface, which decreases the interfacial tension andtherefore facilitates droplet disruption. Thus, the quicker the emulsifier adsorbs to the surfaceof the droplets during homogenization, the smaller the droplets produced. Second, the abilityof emulsifiers to enhance the interfacial rheology of emulsion droplets hampers the breakupof droplets, which leads to larger droplet sizes than those expected from the equilibriuminterfacial tension (Lucassen-Reyanders and Kuijpers 1992, Janssen et al. 1994, Williams etal. 1997). These two effects partly account for the poor correlation between droplet size andequilibrium interfacial tension reported in the literature (Walstra 1983).

6.3.2. Droplet Coalescence

Emulsions are highly dynamic systems in which the droplets continuously move around andfrequently collide with each other (Chapter 7). Droplet–droplet collisions are particularlyrapid during homogenization because of the intense mechanical agitation of the emulsion. Ifdroplets are not protected by a sufficiently strong emulsifier membrane, they tend to coalescewith one another during a collision (Walstra 1993b). Immediately after the disruption of anemulsion droplet, there is insufficient emulsifier present to completely cover the newlyformed surface, and therefore the new droplets are more likely to coalescence with theirneighbors. To prevent coalescence, it is necessary to form a sufficiently concentrated emul-sifier membrane around the droplets before they have time to collide with their neighbors.The size of the droplets produced during homogenization therefore depends on the time takenfor the emulsifier to be adsorbed to the surface of the droplets (τadsorption) relative to the timebetween droplet–droplet collisions (τcollision). These times depend on the flow profile that thedroplets experience, as well as the nature of the emulsifier used. Estimates of the adsorptionand collision times have been established by Walstra (1983) for laminar and turbulent flowconditions (Table 6.1).

The equations given in Table 6.1 are only strictly applicable to dilute emulsions; neverthe-less, they do give some useful insights into the factors which influence homogenization.Ideally, a food manufacturer wants to minimize droplet coalescence during homogenizationby ensuring that τadsorption/τcollision << 1. For both laminar and turbulent flow conditions, thisratio decreases as the dispersed-phase volume fraction decreases, the size of the dropletsincreases, the excess surface concentration decreases, and the concentration of emulsifierincreases. Under laminar flow conditions, the shear rate has no effect on τadsorption/τcollision,because the decrease in adsorption time due to increasing G is counterbalanced by thedecrease in collision time. Under turbulent flow conditions, increasing the power inputdecreases τadsorption/τcollision, because the decrease in adsorption time is greater than the de-crease in the collision time. The above equations assume that every encounter between anemulsifier and droplet leads to adsorption and that every encounter between two droplets


TABLE 6.1Equations for Calculating the Adsorption andCollision Times of Droplets in Emulsions UnderLaminar and Turbulent Flow Conditions

Laminar flow Turbulent flow

τabsorption ≈ 20Γdm Gc

τηεabsorption ≈ 10Γ

dmc

C

τ πφcollision =

8Gτ

φρεcollision = 1

15

23 d C

Note: Γ is the excess surface concentration of the emulsifier, G is theshear rate, ε is the power density, ρC is the density of the contin-uous phase, d is the droplet diameter, φ is the dispersed-phasevolume fraction, γ is the interfacial tension, and mc is the concen-tration of the emulsifier (mol m–3).

Source: Walstra, 1983, 1993.

leads to coalescence. In practice, these mechanisms are not usually 100% efficient becauseof repulsive hydrodynamic and colloidal interactions (Hunter 1986), as well as the Gibbs–Marangoni effect (Walstra 1993b).

The importance of emulsifier adsorption kinetics on the size of the droplets producedduring homogenization has been demonstrated experimentally (Schubert and Armbruster1992). Under the same homogenization conditions, it has been shown that emulsifiers whichadsorb rapidly produce smaller droplet sizes than those which adsorb slowly. Most foodemulsifiers do not adsorb rapidly enough to completely prevent droplet coalescence, and sothe droplet size achieved during homogenization is greater than that which is theoreticallypossible (Stang et al. 1994).

In addition to the factors already mentioned, the tendency for droplets to coalesce during(or shortly after) homogenization depends on the effectiveness of the interfacial membranein resisting coalescence during a droplet–droplet encounter. The resistance of an interfacialmembrane to coalescence depends on the concentration of emulsifier molecules present, aswell as their structural and physicochemical properties (e.g., dimensions, electrical charge,packing, and interactions) (Chapters 4 and 7).

6.3.3. The Role of the Emulsifier

The above discussion has highlighted two of the most important functions of emulsifiersduring the homogenization process:

1. They decrease the interfacial tension between the oil and water phases, therebyreducing the amount of energy required to deform and disrupt the droplets.

2. They form a protective coating around the droplets which prevents them fromcoalescing with each other.

The size of the droplets produced during homogenization therefore depends on a numberof different characteristics of an emulsifier: (1) the ratio of emulsifier to dispersed phase —there must be sufficient emulsifier present to completely cover the surfaces of the droplets


formed; (2) the time required for the emulsifier to move from the bulk phase to the dropletsurface — the faster the adsorption time, the smaller the droplet size; (3) the probability thatan emulsifier molecule will be adsorbed to the surface of a droplet during an encounterbetween it and the droplet — the greater the adsorption efficiency, the smaller the dropletsize; (4) the amount that the emulsifier reduces the interfacial tension — the greater theamount, the smaller the droplet size; and (5) the effectiveness of the emulsifier membrane inprotecting the droplets against coalescence — the better the protection, the smaller the dropletsize.

6.4. HOMOGENIZATION DEVICES

A number of different types of homogenization device have been developed to produce foodemulsions. Each of these devices has its own advantages and disadvantages and range ofmaterials where it is most suitably applied. The choice of a particular homogenizer dependson whether the emulsion is being made in a factory or in a laboratory, the equipmentavailable, the volume of material to be homogenized, the throughput, the nature of the startingmaterials, the desired droplet size distribution, the required physicochemical properties of thefinal product, and the cost of purchasing and running the equipment. The most importanttypes of homogenizer used in the food industry are discussed below.

6.4.1. High-Speed Blenders

High-speed blenders are the most commonly used method for directly homogenizing oil andaqueous phases in the food industry (Brennan et al. 1981, Loncin and Merson 1979, Fellows1988). The liquids to be homogenized are placed in a suitable vessel (Figure 6.5), which maycontain as little as a few cubic centimeters or as much as several cubic meters of liquid, andare then agitated by a stirrer which rotates at high speeds (typically 20 to 2000 rev min–1).The rapid rotation of the blade generates a combination of longitudinal, rotational, and radial

FIGURE 6.5 High-speed blenders are often used in the food industry to directly homogenize oil andaqueous phases.


TABLE 6.2Comparison of Different Types of Homogenizer

Relative energy Minimum SampleThroughput efficiency droplet size viscosity

High-speed blender Batch Low 2 µm Low to mediumColloid mill Continuous Intermediate 1 µm Medium to highHigh-pressure homogenizer Continuous High 0.1 µm Low to mediumUltrasonic probe Batch Low 0.1 µm Low to mediumUltrasonic jet homogenizer Continuous High 1 µm Low to mediumMicrofluidization Continuous High <0.1 µm Low to mediumMembrane processing Batch or continuous High 0.3 µm Low to medium

velocity gradients in the liquids which disrupts the interface between the oil and water,causes the liquids to become intermingled, and breaks the larger droplets into smaller ones(Fellows 1988). Efficient homogenization is achieved when the horizontal and vertical flowprofiles distribute the liquids evenly throughout the vessel, which can be facilitated by fixingbaffles to the inside walls of the vessel (Gopal 1968). The design of the stirrer alsodetermines the efficiency of the homogenization process; a number of different types areavailable for different situations (e.g., blades, propellers, and turbines) (Fellows 1988).Blending generally leads to an increase in the temperature of an emulsion because some ofthe mechanical energy is converted into heat due to viscous dissipation. If any of theingredients in the emulsion are sensitive to heat, it may be necessary to control the tempera-ture of the vessel during homogenization. High-speed blenders are particularly useful forpreparing emulsions with low or intermediate viscosities (Table 6.2). The droplet sizeusually decreases as the homogenization time or the rotation speed of the stirrer is increased,until a lower limit is achieved which depends on the nature and concentration of theingredients used. Typically, the droplets produced by a high-speed blender range betweenabout 2 and 10 µm in diameter.

6.4.2. Colloid Mills

Colloid mills are widely used in the food industry to homogenize medium- and high-viscosity liquids (Loncin and Merson 1979, Walstra 1983, Fellows 1988). A variety ofdifferent designs of colloid mill are commercially available, but they all operate on fairlysimilar physical principles (Figure 6.6). The liquids to be homogenized are fed into thecolloid mill in the form of a coarse emulsion, rather than as separate oil and aqueous phases,because the device is much more efficient at reducing the size of the droplets in a preexistingemulsion (secondary homogenization) than at homogenizing two separate phases (primaryhomogenization). The coarse emulsion is usually produced directly from the oil and aqueousphases using a high-speed blender. It is then fed into the homogenizer and flows through anarrow gap between two disks: the rotor (a rotating disk) and the stator (a static disk). Therapid rotation of the rotor generates a shear stress in the gap which causes the larger dropletsto be broken down into smaller ones. The intensity of the shear stresses can be altered byvarying the thickness of the gap between the rotor and stator (from about 50 to 1000 µm),varying the rotation speed (from about 1000 to 20,000 rev min–1), or by using disks that haveroughened surfaces or interlocking teeth (Gopal 1968). Droplet disruption can also be en-hanced by increasing the length of time the emulsion spends in the colloid mill, either bydecreasing the flow rate or by passing the emulsion through the device a number of times.Typically, the flow rate can be varied between about 4 and 20,000 l h–1. It should be noted


FIGURE 6.6 Colloid mills are mainly used in the food industry to homogenize intermediate- andhigh-viscosity materials.

that many of the factors which increase the effectiveness of droplet disruption also increasethe manufacturing costs (by increasing energy costs or reducing product throughput). Foodmanufacturers must therefore select the rotation speed, gap thickness, rotor/stator type, andthroughput which give the best compromise between droplet size and manufacturing costs.It is usually necessary to have some form of cooling device as part of a colloid mill in orderto offset the increase in temperature caused by viscous dissipation losses. Colloid mills aremore suitable for homogenizing intermediate- and high-viscosity fluids (such as peanutbutter, fish, or meat pastes) than high-pressure valve or ultrasonic homogenizers (Table 6.2).Typically, they can be used to produce emulsions with droplet diameters between about 1 and5 µm.

6.4.3. High-Pressure Valve Homogenizers

High-pressure valve homogenizers are the most commonly used method of producing fineemulsions in the food industry. Like colloid mills, they are more effective at reducing the sizeof the droplets in a preexisting emulsion than at creating an emulsion directly from twoseparate liquids (Pandolfe 1991, 1995). A coarse emulsion is usually produced using a high-speed blender and is then fed directly into the input of the high-pressure valve homogenizer.The homogenizer has a pump which pulls the coarse emulsion into a chamber on its back-stroke and then forces it through a narrow valve at the end of the chamber on its forwardstroke (Figure 6.7). As the coarse emulsion passes through the valve, it experiences acombination of intense shear, cavitational, and turbulent flow conditions which cause thelarger droplets to be broken down into smaller ones (Phipps 1985). A variety of differenttypes of valve have been designed for different types of application. Most commercialhomogenizers use spring-loaded valves so that the gap through which the emulsion passes canbe varied (typically between about 15 and 300 µm). Decreasing the gap size increases thepressure drop across the valve, which causes a greater degree of droplet disruption andsmaller droplets to be produced. On the other hand, narrowing the gap increases the energyinput required to form an emulsion, thereby increasing manufacturing costs. Experimentshave shown that there is an approximately linear relationship between the logarithm of thehomogenization pressure (P) and the logarithm of the droplet diameter (d) (i.e., log d ∝ logP) (Walstra 1983, Phipps 1985). Thus a food manufacturer is able to develop empiricalequations which can be used to predict the homogenization pressure required to produce anemulsion with a given droplet size. The throughputs of industrial homogenizers typically vary


between about 100 to 20,000 l h–1, whereas homogenization pressures vary between about 3and 20 MPa.

Some commercial devices use a “two-stage” homogenization process, in which the emul-sion is forced through two consecutive valves. The first valve is at high pressure and isresponsible for breaking up the droplets, whereas the second valve is at low pressure and isresponsible for disrupting any “flocs” which are formed during the first stage (Phipps 1985).

High-pressure valve homogenizers can be used to produce a wide variety of different foodproducts, although they are most suitable for low- and intermediate-viscosity materials,particularly when a small droplet size is required. If the oil and aqueous phases have beenblended prior to homogenization, it is often possible to create an emulsion with submicronparticles using a single pass through the homogenizer (Pandolfe 1995). If very fine emulsiondroplets are required, it is usually necessary to pass the emulsion through the homogenizera number of times. Emulsion droplets with diameters as small as 0.1 µm can be producedusing this method. The temperature rise in a high-pressure valve homogenizer is usually fairlysmall, but it can become appreciable if the emulsion is recirculated or extremely highpressures are used. In these cases, it may be necessary to keep the emulsion cool by usinga water-jacketed homogenization chamber.

6.4.4. Ultrasonic Homogenizers

The possibility of using ultrasound to form emulsions was realized in the early part of thiscentury, and ultrasonic homogenizers have been widely used for this purpose in both industryand research since then (McCarthy 1964, Gopal 1968). This type of homogenizer utilizeshigh-intensity ultrasonic waves which generate intense shear and pressure gradients within amaterial that disrupt the droplets mainly due to cavitational effects (Section 6.3.1.2).

A number of methods are available for generating high-intensity ultrasonic waves, but onlytwo are commonly used in the food industry: piezoelectric transducers and liquid jet genera-tors (Gopal 1968). Piezoelectric transducers are used in the benchtop ultrasonic homogenizersthat are found in many research laboratories. They are ideal for preparing small volumes ofemulsion (a few to a few hundred cubic centimeters), which is an important considerationin research laboratories because the chemicals used are often expensive. An ultrasonictransducer consists of a piezoelectric crystal contained within a protective metal casing,

FIGURE 6.7 High-pressure valve homogenizers are used to produce emulsions with fine dropletsizes.


FIGURE 6.8 Liquid jet generators are commonly used in the food industry for the continuousproduction of emulsions.

which is usually tapered at the end. A high-intensity electrical wave is applied to thetransducer, which causes the piezoelectric crystal to rapidly oscillate and generate an ultra-sonic wave. The ultrasonic wave is directed toward the tip of the transducer, where it radiatesinto the surrounding liquids and generates intense pressure and shear gradients (mainly dueto cavitational effects) which cause the liquids to be broken up into smaller fragments andintermingle with one another. The fact that the ultrasonic energy is focused on a small volumeof the sample near the tip of the ultrasonic transducer means that it is important to have goodagitation in the sample container. In small vessels, this is achieved by the fluid flow inducedby the ultrasonic field itself, but in large vessels, it is often necessary to have additionalagitation to ensure effective mixing and homogenization of all of the sample. To create astable emulsion, it is usually necessary to irradiate a sample with ultrasound for periodsranging from a few seconds to a few minutes. Continuous application of ultrasound to asample can cause appreciable heating, and so it is often advantageous to apply the ultrasoundin a number of short bursts.

Ultrasonic jet homogenizers are used mainly for the industrial preparation of food emul-sions (Figure 6.8). A stream of fluid is made to impinge on a sharp-edged blade, whichcauses the blade to rapidly vibrate, thus generating an intense ultrasonic field that breaks upany droplets in its immediate vicinity due to a combination of cavitation, shear, and turbu-lence (Gopal 1968). The major advantages of this device are that it can be used for thecontinuous production of emulsions, it can generate very small droplets, and it is moreenergy efficient than high-pressure valve homogenizers (i.e., less energy is required toproduce droplets of the same size). Even so, the vibrating blade is prone to erosion becauseof the ultrasonic field, which means that it has to be replaced frequently. Fluid flow ratesbetween 1 and 5000 l min–1 are possible using this technique.

The principal factors determining the efficiency of ultrasonic homogenizers are the inten-sity, duration, and frequency of the ultrasonic waves (Gopal 1968). Below a frequency ofabout 16 kHz, ultrasonic waves become audible and are therefore objectionable to users.Emulsions can be formed using ultrasonic waves with frequencies as high as 5 MHz, but thehomogenization efficiency decreases with increasing frequency. For these reasons, mostcommercial devices use ultrasonic waves with frequencies between about 20 and 50 kHz. Thesize of the droplets produced during homogenization can be decreased by increasing eitherthe intensity of the ultrasonic radiation or the length of time it is applied.

6.4.5. Microfluidization

Microfluidization is a technique which is capable of creating emulsions with extremely smalldroplet sizes directly from the individual oil and aqueous phases (Dickinson and Stainsby1988). Separate streams of the oil and aqueous phases are accelerated to a high velocity and


FIGURE 6.10 Batch version of a membrane homogenizer.

FIGURE 6.9 In a microfluidizer, the oil and aqueous phases are brought together at a high velocity,which causes the intermingling of the liquids and the disruption of the droplets.

then made to simultaneously impinge on a surface, which causes them to be intermingledwith each other and broken up into droplets (Figure 6.9). Extremely small emulsion dropletscan be produced by recirculating an emulsion through the microfluidizer a number of times(Strawbridge et al. 1995). The jet homogenizer developed by Dickinson and co-workers is anexample of a laboratory-scale instrument based on the principle of microfluidization whichcan be used to create emulsions when the volume of starting materials is limited or expensive(Dickinson and Stainsby 1988).

6.4.6. Membrane Homogenizers

An emulsion is formed by forcing one immiscible liquid into another through a glassmembrane which contains a uniform pore size (Figure 6.10). The size of the droplets formed


depends on the diameter of the pores in the membrane and the interfacial tensionsbetween the oil and water phases (Kandori 1995). Membranes can be manufactured withdifferent pore diameters so that emulsions with different droplet sizes can be produced(Kandori 1995). The membrane must be sufficiently strong so that it is not broken by thepressures applied to the fluids during homogenization. The membrane technique can be usedas either a batch or a continuous process. In the batch process, droplets are formed byforcing the dispersed phase through a cylindrical membrane which is dipped into a vesselcontaining the continuous phase (Figure 6.10). In the continuous process, the homogenizerconsists of a cylindrical membrane through which the continuous phase flows, which islocated within a tube through which the dispersed phase flows. The dispersed phase ispressurized so that it is forced through the membrane, where it forms small droplets in thecontinuous phase.

An increasing number of applications for the membrane homogenizer are being identified,and the technique can now be purchased commercially for preparing emulsions in thelaboratory (Kandori 1995). These instruments can be used to produce oil-in-water, water-in-oil, and multiple emulsions. The major advantages of membrane homogenizers are theirability to produce emulsions with very narrow droplet size distributions and the fact that theyare highly energy efficient because much less energy is lost due to viscous dissipation.Emulsions with mean droplet sizes between about 0.3 and 10 µm can be produced using thistechnique.

6.4.7. Homogenization Efficiency

The energy efficiency of a homogenizer (EH) can be calculated by comparing the minimumamount of energy theoretically required to form an emulsion (∆Emin) with the actual amountof energy that is expended during homogenization (∆Etotal):

EE

EH = ×∆∆

min

total

100 (6.6)

The minimum amount of energy required to form an emulsion is equal to that needed toincrease the interfacial area between the oil and water phases: ∆Emin = ∆Aγ, where ∆A is theincrease in interfacial area and γ is the interfacial tension. For a typical oil-in-water emulsion,∆Emin has a value of about 3 kJ m–3, assuming that φ = 0.1, r = 1 µm, and γ = 10 mN m–1

(Walstra 1983). The actual amount of energy required to form an emulsion depends on thetype of homogenizer used and the operating conditions. For a high-pressure valve homog-enizer, ∆Etotal is typically about 10,000 kJ m–3, and so the homogenization efficiency is lessthan 0.1% (Walstra 1983). The reason that homogenization is such an inefficient process isbecause the disruption of small droplets requires the generation of extremely high pressuregradients. These pressure gradients must be large enough to overcome the Laplace pressuregradient (≈2γ/r2), which is about 2 × 1010 Pa m–1 for a 1-µm emulsion droplet. The pressuregradient due to shear forces acting across a droplet is given by GηC/r, which indicates thatthe shear rate must exceed about 2 × 107 s–1 in order to disrupt the droplets. The movementof liquids at such high shear rates leads to large amounts of energy dissipation because offrictional losses. The conversion of mechanical work into heat also accounts for the increasein temperature observed during homogenization. To improve the energy efficiency of ahomogenizer, it is necessary to either reduce the pressure gradient required to disrupt thedroplets (e.g., by reducing the interfacial tension) or minimize the viscous dissipation causedby fluid flow.


6.4.8. Comparison of Homogenizers

The choice of a homogenizer for a particular application depends on a number of factors,including the volume of sample to be homogenized, the desired throughput, energy consump-tion, the physicochemical properties of the component phases, the desired droplet size dis-tribution, the equipment available, initial costs, and running costs. After choosing the mostsuitable type of homogenizer, one must select the optimum operating conditions for thatparticular device, such as flow rate, pressure, gap thickness, temperature, homogenizationtime, and rotation speed.

The conversion of separate oil and aqueous phases into an emulsion containing fairly largedroplets (>2 µm) is normally achieved using a high-speed blender. For industrial applicationsusing high-viscosity fluids (0.1 < ηC < 1 Pa · s), a colloid mill is the most efficient type ofhomogenizer, but for lower viscosity liquids, either the high-pressure valve homogenizer orultrasonic jet homogenizer is more suitable. For fundamental studies, one often needs smallsample volumes; therefore, specially designed laboratory homogenizers are available, whichare either scaled-down versions of the industrial equipment or instruments specifically de-signed for use in the laboratory. For studies where the ingredients are limited or expensive,an ultrasonic transducer could be used because it can produce small sample sizes. For studieswhere it is important to have monodisperse emulsions, it may be advantageous to use amembrane homogenizer.

The selection of an appropriate homogenizer for a particular application usually involvesclose cooperation between the food processor and the manufacturer of the equipment. Thefood processor must first specify the desired throughput, pressure, temperature, particle size,hygiene requirements, and properties of the sample. The equipment manufacturer will thenbe able to recommend a piece of equipment that is most suitable for the specific product (e.g.,size, valve design, flow rates, and construction materials). It is good practice for foodprocessors to test a number of homogenizers from different manufacturers under the condi-tions that will be used in the factory prior to making a purchase.

6.5. FACTORS WHICH INFLUENCE DROPLET SIZE

The size of the droplets produced during homogenization is important because it determinesthe stability, appearance, and texture of the final product. To create a product with specificproperties, it is therefore necessary to ensure that the majority of the droplets fall within somepreestablished size range. For this reason, it is important for food scientists to be aware ofthe major factors which influence droplet size.

6.5.1. Emulsifier Type and Concentration

For a fixed concentration of oil, water, and emulsifier, there is a maximum interfacial areawhich can be completely covered by the emulsifier. As homogenization proceeds, the size ofthe droplets decreases and the interfacial area increases. Once the droplets fall below a certainsize, there is insufficient emulsifier present to completely cover their surface, and so they tendto coalesce with their neighbors. The minimum size of stable droplets that can be producedduring homogenization (assuming monodisperse droplets*) is therefore governed by the typeand concentration of emulsifier present:

rcS

min =⋅ ⋅3 Γsat φ

(6.7)

* For a polydisperse emulsion, the radius used in Equation 6.6 should be the volume–surface mean radius(Chapter 1).


where Γsat is the excess surface concentration of the emulsifier at saturation (in kg m–2), φ isthe dispersed-phase volume fraction, and cS is the concentration of emulsifier in the emulsion(in kg m–3). Equation 6.7 indicates that the minimum droplet size can be decreased byincreasing the emulsifier concentration, decreasing the droplet concentration, or using anemulsifier with a smaller Γsat. For a 10% oil-in-water emulsion containing 1% of emulsifier,the minimum droplet radius is about 60 nm (assuming Γsat = 2 × 10–6 kg m–2). In practice,there are a number of factors which mean that the droplet size produced during homogeni-zation is greater than the theoretical minimum.

In order to attain the theoretical minimum droplet size, a homogenizer must be capableof generating a pressure gradient that is large enough to disrupt any droplets that are greaterthan rmin (i.e., >2γ/rmin

2) (Section 6.4.7). Some types of homogenizer are not capable ofgenerating such high pressure gradients and are therefore not suitable for producing emul-sions with small droplet sizes, even though there may be sufficient emulsifier present(Walstra 1983). The emulsion must also spend sufficient time within the homogenizationzone for all of the droplets to be completely disrupted. If an emulsion passes through ahomogenizer too rapidly, some of the droplets may not be disrupted. Even if a homogenizeris capable of producing small droplets, the emulsifier molecules must adsorb rapidly enoughto form a protective interfacial layer around the droplets which prevents them from coalesc-ing with their neighbors.

The emulsifier also influences the droplet size by reducing the interfacial tension betweenthe oil and aqueous phases and therefore facilitating droplet disruption. Consequently, themore rapidly an emulsifier adsorbs, and the greater the reduction of the interfacial tension,the smaller the droplets that can be produced at a certain energy input.

Many different types of emulsifier can be used in the food industry, and each of theseexhibits different characteristics during homogenization (e.g., the speed at which they adsorb,the maximum reduction in interfacial tension, and the effectiveness of the interfacial mem-brane in preventing droplet coalescence). Factors which influence the adsorption kinetics ofemulsifiers, and their effectiveness at reducing interfacial tension, were discussed in Chapter5, and factors which affect the stability of droplets against coalescence will be covered inChapter 7. A food manufacturer must select the most appropriate emulsifier for each type offood product, taking into account its performance during homogenization, solution condi-tions, cost, availability, legal status, ability to provide long-term stability, and the desiredphysicochemical properties of the product.

6.5.2. Energy Input

The size of the droplets in an emulsion can be reduced by increasing the amount of energysupplied during homogenization (as long as there is sufficient emulsifier to cover the surfacesof the droplets formed). The energy input can be increased in a number of different waysdepending on the nature of the homogenizer. In a high-speed blender, the energy input canbe enhanced by increasing the rotation speed or the length of time that the sample is blended.In a high-pressure valve homogenizer, it can be enhanced by increasing the homogenizationpressure or recirculating the emulsion through the device. In a colloid mill, it can be enhancedby using a narrower gap between the stator and rotator, increasing the rotation speed, by usingdisks with roughened surfaces, or by passing the emulsion through the device a number oftimes. In an ultrasonic homogenizer, the energy input can be enhanced by increasing theintensity of the ultrasonic wave or by sonicating for a longer time. In a microfluidizer, theenergy input can be enhanced by increasing the velocity at which the liquids are brought intocontact with each other or by recirculating the emulsion. In a membrane homogenizer, theenergy input can be enhanced by increasing the pressure at which the liquid is forced throughthe membrane. Under a given set of homogenization conditions (energy input, temperature,


composition), there is a certain size below which the emulsion droplets cannot be reducedwith repeated homogenization, and therefore homogenizing the system any longer would beinefficient.

Increasing the energy input usually leads to an increase in manufacturing costs, andtherefore a food manufacturer must establish the optimum compromise between droplet size,time, and cost. The energy input required to produce an emulsion containing droplets of agiven size depends on the energy efficiency of the homogenizer used (Walstra 1983).

Under most circumstances, there is a decrease in droplet size as the energy input isincreased. Nevertheless, there may be occasions when increasing the energy actually leadsto an increase in droplet size because the effectiveness of the emulsifier is reduced byexcessive heating or exposure to high pressures. This could be particularly important forprotein-stabilized emulsions, because the molecular structure and functional properties ofproteins are particularly sensitive to changes in their environmental conditions. For example,globular proteins, such as β-lactoglobulin, are known to unfold and aggregate when they areheated above a certain temperature, which reduces their ability to stabilize emulsions (Sec-tion 4.6).

6.5.3. Properties of Component Phases

The composition and physicochemical properties of both the oil and aqueous phases influ-ence the size of the droplets produced during homogenization (Phipps 1985). Variations inthe type of oil or aqueous phase will alter the viscosity ratio (ηD/ηC) which determines theminimum size that can be produced under steady-state conditions (Section 6.3). Different oilshave different interfacial tensions when placed in contact with water because they havedifferent molecular structures or because they contain different amounts of surface-activeimpurities, such as free fatty acids, monoacylglycerols, or diacylglycerols. These surface-active lipid components tend to accumulate at the oil–water interface and lower the interfacialtension, thus lowering the amount of energy required to disrupt a droplet.

The aqueous phase of an emulsion may contain a wide variety of components, includingminerals, acids, bases, biopolymers, sugars, alcohols, ice crystals, and gas bubbles. Many ofthese components will alter the size of the droplets produced during homogenizationbecause of their influence on rheology, interfacial tension, coalescence stability, or ad-sorption kinetics. For example, the presence of low concentrations of short-chain alcoholsin the aqueous phase of an emulsion reduces the size of the droplets produced duringhomogenization because of the reduction in interfacial tension (Banks and Muir 1988). Thepresence of biopolymers in an aqueous phase has been shown to increase the droplet sizeproduced during homogenization due to their ability to suppress the formation of smalleddies during turbulence (Walstra 1983). Protein-stabilized emulsions cannot be producedclose to the isoelectric point of a protein or at high electrolyte concentrations because theproteins are susceptible to aggregation. A knowledge of the composition of both the oil andaqueous phases of an emulsion and the role that each component plays during homogeni-zation is therefore important when optimizing the size of the droplets produced by ahomogenizer.

Experiments have shown that the smallest droplet size that can be achieved using a high-pressure valve homogenizer increases as the dispersed-phase volume fraction increases (Phipps1985). There are a number of possible reasons for this: (1) increasing the viscosity of anemulsion may suppress the formation of eddies responsible for breaking up droplets; (2) ifthe emulsifier concentration is kept constant, there may be an insufficient amount present tocompletely cover the droplets; and (3) the rate of droplet coalescence is increased.


6.5.4. Temperature

Temperature influences the size of the droplets produced during homogenization in a numberof ways. The viscosity of both the oil and aqueous phases is temperature dependent, andtherefore the minimum droplet size that can be produced may be altered because of avariation in the viscosity ratio (ηD/ηC) (Section 6.3). Heating an emulsion usually causes aslight reduction in the interfacial tension between the oil and water phases, which would beexpected to facilitate the production of small droplets (Equation 6.1). Certain types ofemulsifiers lose their ability to stabilize emulsion droplets against aggregation when they areheated above a certain temperature. For example, when small-molecule surfactants are heatedclose to their phase inversion temperature, they are no longer effective in preventing dropletcoalescence, or when globular proteins are heated above a critical temperature, they unfoldand aggregate (Chapter 4). Alterations in temperature also influence the competitive adsorp-tion of surface-active components, thereby altering interfacial composition (Dickinson andHong 1994).

The temperature is also important because it determines the physical state of the lipid phase(Section 4.2). It is practically impossible to homogenize a fat that is either completely orsubstantially solid because it will not flow through a homogenizer or because of the hugeamount of energy required to break up the fat crystals into small particles. There are alsoproblems associated with the homogenization of oils that contain even small amounts of fatcrystals because of partial coalescence (Chapter 7). The crystals from one droplet penetrateanother droplet, leading to the formation of a “clump.” Extensive clump formation leads tothe generation of large particles and to a dramatic increase in the viscosity, which wouldcause a homogenizer to become blocked. For this reason, it is usually necessary to warm asample prior to homogenization to ensure that the lipid phase is completely liquid. Forexample, milk fat is usually heated to about 40°C to melt all the crystals prior to homogeni-zation (Phipps 1985).

6.6. DEMULSIFICATION

Demulsification is the process whereby an emulsion is converted into the separate oil andaqueous phases from which it was comprised and is therefore the opposite of homogeniza-tion (Lissant 1983, Menon and Wasan 1985, Hunter 1989). Demulsification is important ina number of technological processes in the food industry (e.g., oil recovery or the separationof lipid and aqueous phases). Demulsification is also important in research and developmentbecause it is often necessary to divide an emulsion into the separate oil and aqueous phasesso that their composition or properties can be characterized. For example, the oil phase mustoften be extracted from an emulsion in order to determine the extent of lipid oxidation(Coupland et al. 1996) or to measure the partition coefficient of a food additive (Huang etal. 1997). Demulsification is achieved by causing the droplets to come into close contactwith each other and then to coalesce. As this process continues, it eventually leads to thecomplete separation of the oil and aqueous phases. A knowledge of the physical principlesof demulsification requires an understanding of the factors that determine the stability ofemulsions to flocculation and coalescence (Chapter 7).

A variety of different types of emulsifier are used in the food industry to stabilize dropletsagainst flocculation and coalescence (Chapter 4). Each type of emulsifier relies on differentphysicochemical mechanisms to prevent droplet aggregation, including electrostatic, steric,hydration, and thermal fluctuation interactions (Chapter 3). The selection of the most appro-priate demulsification technique for a given emulsion therefore depends on a knowledge of


the type of emulsifier used to stabilize the system and the mechanisms by which it providesstability.

6.6.1. Nonionic Surfactants

Nonionic surfactants usually stabilize emulsion droplets against flocculation through a com-bination of polymeric steric, hydration, and thermal fluctuation interactions (Section 4.5).Nevertheless, the interfacial membranes formed by nonionic surfactants are often unstable torupture when the droplets are brought into close contact. Demulsification can therefore beachieved by altering the properties of an emulsion so that the droplets come into close contactfor prolonged periods. This can often be achieved by heating an emulsion so that the polarhead groups of the surfactant molecules become dehydrated, because this reduces the hydra-tion repulsion between the droplets and allows them to come closer together. In addition, theoptimum curvature of the surfactant monolayer tends toward zero as the size of the headgroup decreases, which increases the likelihood of coalescence (Section 4.5). Thisdemulsification technique cannot be used for some emulsions stabilized by nonionic surfac-tants because the phase inversion temperature is much greater than 100°C or because heatingmay cause degradation or loss of one of the components being analyzed. In these cases, it isnecessary to induce demulsification using alternative methods.

The addition of medium-chain alcohols has also been found to be effective in promotingdemulsification in some systems (Menon and Wasan 1985). There are two possible explana-tions for this behavior: (1) the alcohol displaces some of the surfactant molecules from theinterface and forms an interfacial membrane which provides little protection against dropletaggregation or (2) the alcohol molecules are able to get between the tails of the surfactantmolecules at the interface, thereby increasing the optimum curvature of the interface towardzero and increasing the likelihood of coalescence (Section 4.5). In some emulsions, it ispossible to promote droplet coalescence by adding a strong acid which cleaves the headgroups of the surfactants from their tails so that the polar head group moves into the aqueousphase and the nonpolar tail moves into the droplet, thereby providing little protection againstdroplet coalescence.

6.6.2. Ionic Surfactants

Ionic surfactants stabilize droplets against coalescence principally by electrostatic repulsion(Chapter 3). Like nonionic surfactants, the membranes formed by ionic surfactants are notparticularly resistant to rupture once the droplets are brought into close contact (Evans andWennerstrom 1994). The most effective method of inducing droplet coalescence in thesesystems is therefore to reduce the magnitude of the electrostatic repulsion between thedroplets (Menon and Wasan 1985). This can be achieved by adding electrolyte to theaqueous phase of the emulsion so as to screen the electrostatic interactions. Sufficientelectrolyte must be added so that the energy barrier between the droplets becomes of theorder of the thermal energy or less (Chapter 3). This process can most easily be achievedusing multivalent ions, because they are more effective at screening electrostatic interactionsat low concentrations than monovalent ions. Alternatively, the pH may be altered so that thesurfactant loses its charge, which depends on the dissociation constant of the ionizablegroups. Electromechanical methods can also be used to promote demulsification. An electricfield is applied across an emulsion, which causes the charged droplets to move toward theoppositely charged electrode. A semipermeable membrane is placed across the path of thedroplets, which captures the droplets but allows the continuous phase to pass through. Thedroplets are therefore forced against the membrane until their interfacial membranes areruptured and they coalesce.


6.6.3. Biopolymer Emulsifiers

Biopolymers principally stabilize droplets against coalescence through a combination ofelectrostatic and polymeric steric interactions. In addition, they tend to form thick viscoelasticmembranes that are highly resistant to rupture. There are two different strategies that can beuse to induce droplet coalescence in this type of system:

1. The biopolymer can be digested by strong acids or enzymes so that it is broken intosmall fragments that are either not surface active or do not form a sufficientlystrong membrane.

2. The biopolymers are displaced from the interface by small-molecule surfactants,and then the droplets are destabilized using one of the methods described above.Some proteins are capable of forming an interfacial membrane in which the mol-ecules are covalently bound to each other through disulfide bonds. In order todisplace these proteins, it may be necessary to cleave the disulfide bonds prior todisplacing the proteins (e.g., by adding mercaptoethanol).

The Gerber and Babcock methods of determining the total fat content of milk are examplesof the first of these strategies, while the detergent method is an example of the second (Pike1994).

6.6.4. General Methods of Demulsification

A variety of physical techniques are available which can be used to promote demulsificationin most types of emulsions. In all of the demulsification processes mentioned above, theseparation of the oil phase from the aqueous phase can be facilitated by centrifuging theemulsion after the coalescence process has been initiated. In some emulsions, it is alsopossible to separate the phases directly by centrifugation at high speeds, without the need forany pretreatment (Menon and Wasan 1985). Centrifugation forces the droplets to one end ofthe container, which causes their interfacial membranes to become ruptured and thereforeleads to phase separation.

Demulsification can also be achieved using various types of filtration devices (Menon andWasan 1985). The emulsion is passed through a filter which adsorbs emulsion droplets. Whena number of these adsorbed droplets come into close contact, they merge together to form asingle large droplet which is released back into the aqueous phase. As the emulsion passesthrough the filter, this process continues until eventually the oil and water phases are com-pletely separated from each another.

6.6.5. Selection of Most Appropriate Demulsification Technique

In addition to depending on the type of emulsifier present, the choice of an appropriatedemulsification technique also depends on the sensitivity of the other components in thesystem to the separation process. For example, if one is monitoring lipid oxidation or tryingto determine the concentration of an oil-soluble volatile component, it is inadvisable to usea demulsification technique that requires excessive heating. On the other hand, if the samplecontains a lipid phase that is crystalline, it is usually necessary to warm the sample to atemperature where all the fat melts prior to carrying out the demulsification procedure.

6.7. FUTURE DEVELOPMENTS

Homogenization is an extremely important step in the production of emulsion-based foodproducts. The efficiency of this process has a large impact on the bulk physicochemical and


sensory properties of the final product. The progress that has already been made in identifyingthe factors which influence homogenization was reviewed in this chapter. Nevertheless, agreat deal of research is still required before this process can be fully understood because ofthe inherent complexity of the physicochemical processes involved. The existing theoriesneed to be extended to give a more realistic description of emulsion formation in commercialhomogenization devices. In addition, systematic experiments using well-characterized emul-sions and homogenization devices need to be carried out. Our understanding of this area willalso be advanced due to the valuable insights provided by computer simulations. Computermodels have been developed to predict the size distribution of droplets produced in homog-enizers (Lachaise et al. 1996). These models can take into account the competition betweendroplet disruption and droplet coalescence mechanisms, as well as the influence of emulsifieradsorption kinetics on these processes.

Before ending this chapter, it is important to mention that homogenization is only one stepin the formation of a food emulsion. A number of other processing operations usually comebefore or after homogenization, including chilling, freezing, pasteurization, drying, mixing,churning, and whipping. The quality of the final product is determined by the effect that eachof these processing operations has on the properties of the food. Homogenization efficiencymay be influenced by the effectiveness of any of the preceding processing operations, and itmay alter the effectiveness of any of the following processing operations. Thus it is importantto establish the interrelationship between the various food-processing operations on the finalproperties of a product.


7 Emulsion Stability

7.1. INTRODUCTION

The term “emulsion stability” refers to the ability of an emulsion to resist changes in itsproperties over time: the more stable the emulsion, the more slowly its properties change. Anemulsion may become unstable due to a number of different types of physical and chemicalprocesses.* Physical instability results in an alteration in the spatial distribution or structuralorganization of the molecules, whereas chemical instability results in an alteration in thechemical structure of the molecules. Creaming, flocculation, coalescence, partial coalescence,phase inversion, and Ostwald ripening are examples of physical instability (Dickinson andStainsby 1982, Dickinson 1992, Walstra 1996a,b), whereas oxidation and hydrolysis arecommon examples of chemical instability (Fennema 1996a). In practice, two or more of thesemechanisms may operate in concert. It is therefore important for food scientists to identifythe relative importance of each mechanism, the relationship between them, and the factorswhich influence them, so that effective means of controlling the stability and physicochemi-cal properties of emulsions can be established.

The length of time that an emulsion must remain stable depends on the nature of the foodproduct (Dickinson 1992). Some food emulsions are formed as intermediate steps during amanufacturing process and therefore only need to remain stable for a few seconds, minutes,or hours (e.g., cake batter, ice cream mix, margarine premix), whereas others must remainstable for days, months, or even years prior to consumption, (e.g., mayonnaise, salad dress-ings, and cream liqueurs). On the other hand, the production of some foods involves acontrolled destabilization of an emulsion during the manufacturing process (e.g., margarine,butter, whipped cream, and ice cream) (Mulder and Walstra 1974, Boode 1992). One of themajor objectives of emulsion scientists working in the food industry is to establish the factorswhich determine the stability of each type of food product, as well as to elucidate generalprinciples which can be used to predict the behavior of new products or processes. In practice,it is very difficult to quantitatively predict the stability of food emulsions from first principlesbecause of their compositional and structural complexity. Nevertheless, an appreciation of theorigin and nature of the various destabilization mechanisms is still an invaluable tool forcontrolling and improving emulsion stability. Because of the difficulties in theoreticallypredicting emulsion stability, food scientists often rely on the use of analytical techniques toexperimentally monitor changes in emulsion properties over time. By using a combination oftheoretical understanding and experimental measurements, food manufacturers are able topredict the effectiveness of different ingredients and processing conditions on the stabilityand properties of food emulsions.

The rate at which an emulsion breaks down, and the mechanism by which this processoccurs, depends on its composition and microstructure, as well as on the environmentalconditions it experiences during its lifetime (e.g., temperature variations, mechanical agita-

* It should be noted that the properties of emulsions may also change with time due to microbiological changes(e.g., the growth of specific types of bacteria or mold).


tion, and storage conditions). In this chapter, we examine the physicochemical basis of eachof the major destabilization mechanisms, as well as discuss the major factors which influencethem, methods of controlling them, and experimental techniques for monitoring them. Thisinformation is particularly useful for food manufacturers who need to formulate emulsionswith enhanced shelf life or promote emulsion instability in a controlled fashion.

7.2. ENERGETICS OF EMULSION STABILITY

In considering the “stability” of an emulsion, it is extremely important to distinguish betweenits thermodynamic stability and its kinetic stability (Dickinson 1992). Thermodynamics tellsus whether or not a given process will occur, whereas kinetics tells us the rate at which it willproceed if it does occur (Atkins 1994). All food emulsions are thermodynamically unstablesystems and will eventually break down if they are left long enough. For this reason, it isdifferences in kinetic stability which are largely responsible for the diverse range of proper-ties exhibited by different food emulsions.

7.2.1. Thermodynamics

The thermodynamic instability of an emulsion is readily demonstrated if one agitates asealed vessel containing pure oil and pure water and then observes the change in theappearance of the system with time. The optically opaque emulsion which is initially formedby agitation breaks down over time until a layer of oil is observed on top of a layer of water(Figure 6.1).

The origin of this thermodynamic instability can be illustrated by comparing the freeenergy of a system consisting of an oil and an aqueous phase before and after emulsification(Hunter 1989). To simplify this analysis, we will initially assume that the oil and water havesimilar densities so that no creaming or sedimentation occurs. As a consequence, the finalstate consists of a single large droplet suspended in the continuous phase (Figure 7.1), ratherthan a layer of oil on top of a layer of water (see Figure 6.1). In its initial state, prior toemulsification, the free energy is given by:

G G G G TSiOi

Wi

Ii i= + + − config (7.1)

and in its final state, after emulsification, it is given by:

G G G G TSfOf

Wf

If f= + + − config (7.2)

FIGURE 7.1 The formation of an emulsion is thermodynamically unfavorable because of the increasein surface area between the oil and water phases.


where GO, GW, and GI are the free energies of the oil phase, water phase, and the oil–waterinterface, respectively; T is the absolute temperature; and S is the configurational entropy ofthe droplets in the system. The superscripts i and f refer to the initial and final states of thesystem. The free energies of the bulk oil and water phases remain constant before and afterhomogenization: Gi

O = GfO and Gi

W = GfW, and so the difference in free energy between the

initial and final states is given by (Hunter 1989):

∆ ∆ ∆G G G G G TS TS G T Sf iIf

Ii f i

Iformation config config config= − = − − − = −( ) (7.3)

By definition, the difference in interfacial free energy between the initial and final states(∆GI ) is equal to the increase in surface area between the oil and aqueous phases (∆A)multiplied by the interfacial tension (γ): ∆GI = γ∆A. Hence,

∆ ∆ ∆G A T Sformation config= −γ (7.4)

The change in interfacial free energy (γ∆A) is always positive, because the interfacial areaincreases after homogenization, and therefore it opposes emulsion formation. On the otherhand, the configurational entropy term (–T∆Sconfig) is always negative, because the number ofarrangements accessible to the droplets in the emulsified state is much greater than in thenonemulsified state, and therefore it favors emulsion formation. An expression for the con-figurational entropy can be derived from a statistical analysis of the number of configurationsemulsion droplets can adopt in the initial and final states (Hunter 1989):

∆Snk

config = − + − −φ

φ φ φ φ[ ln ( ) ln( )]1 1 (7.5)

where k is Boltzmann’s constant, n is the number of droplets, and φ is the dispersed-phasevolume fraction. In most food emulsions, the configurational entropy is much smaller thanthe interfacial free energy and can be ignored (Hunter 1989). As an example, consider a 10%oil-in-water emulsion containing 1-µm droplets (γ = 0.01 N m–1). The interfacial free energyterm (γ∆A) is about 3 kJ/m3 of emulsion, whereas the configurational entropy term (T∆S) isabout 3 × 10–7 kJ/m3.

The overall free energy change associated with the creation of a food emulsion cantherefore be represented by the following expression:

∆ ∆G Aformation = γ (7.6)

Thus the formation of a food emulsion is always thermodynamically unfavorable, becauseof the increase in interfacial area after emulsification. It should be noted that the configura-tional entropy term can dominate the interfacial free energy term in emulsions in which theinterfacial tension is extremely small and that these systems are therefore thermodynamicallystable (Hunter 1989). This type of thermodynamically stable system is usually referred to asa microemulsion, to distinguish it from thermodynamically unstable (macro)emulsions.

In practice, the oil and water phases normally have different densities, and so it is necessaryto include a free energy term which accounts for gravitational effects (i.e., the tendency forthe liquid with the lowest density to move to the top of the emulsion). This term contributesto the thermodynamic instability of emulsions and accounts for the observed creaming orsedimentation of droplets (Section 7.3.2).


7.2.2. Kinetics

The free energy change associated with emulsion formation determines whether or not anemulsion is thermodynamically stable, but it does not give any indication of the rate at whichthe properties of an emulsion change with time, the type of changes which occur, or thephysical mechanism(s) responsible for these changes. Information about the time dependenceof emulsion stability is particularly important to food scientists who need to create foodproducts which retain their desirable properties for a sufficiently long time under a varietyof different environmental conditions. For this reason, food scientists are usually moreinterested in the kinetic stability of emulsions, rather than their thermodynamic stability.

The importance of kinetic effects can be highlighted by comparing the long-term stabilityof emulsions with the same composition but with different droplet sizes. An emulsion whichcontains small droplets usually has a longer shelf life (greater kinetic stability) than one whichcontains large droplets, even though it is more thermodynamically unstable (because it hasa larger interfacial area, ∆A).

Despite the fact that food emulsions exist in a thermodynamically unstable state, many ofthem remain kinetically stable (metastable) for months or even years. What is the origin ofthis kinetic stability? Conceptually, the kinetic stability of an emulsion can be attributed toan activation energy (∆G*), which must be overcome before it can reach its most thermody-namically favorable state (Figure 7.2). An emulsion which is kinetically stable must have anactivation energy which is significantly larger than the thermal energy of the system (kT). Formost emulsions, an activation energy of about 20 kT is sufficient to provide long-termstability (Friberg 1997). In reality, emulsions have a number of different metastable states,and each of these has its own activation energy. Thus, an emulsion may move from onemetastable state to another before finally reaching the most thermodynamically stable state.A change from one of these metastable states to another may be sufficient to have a delete-rious effect on food quality.

The kinetic stability of emulsions can only be understood with reference to their dynamicnature. The droplets in an emulsion are in a continual state of motion and frequently collideinto one another, due to their Brownian motion, gravity, or applied external forces. Whetherdroplets move apart, remain loosely associated with each other, or fuse together after acollision depends on the nature of the interactions between them. The kinetic stability ofemulsions is therefore largely determined by the dynamics and interactions of the droplets

FIGURE 7.2 Emulsions are thermodynamically unstable systems, but may exist in a metastable stateand therefore be kinetically stable.


FIGURE 7.3 Food emulsions are prone to creaming because of the density difference between the oiland water phases. Inset: the forces acting on an emulsion droplet.

they contain. Consequently, a great deal of this chapter will be concerned with the natureof the interactions between droplets and the factors which determine their movement inemulsions.

Earlier it was mentioned that if pure oil and pure water are agitated together, a temporaryemulsion is formed which rapidly reverts back to its individual components. This is becausethere is a very low activation energy between the emulsified and unemulsified states. Tocreate an emulsion which is kinetically stable for a reasonably long period of time, it isnecessary to have either an emulsifier or a thickening agent present that produces an activa-tion energy that is sufficiently large to prevent instability. An emulsifier adsorbs to thesurface of freshly formed droplets and forms a protective membrane which prevents themfrom merging together, while a thickening agent increases the viscosity of the continuousphase so that droplets collide with one another less frequently (Chapter 4). The role ofemulsifiers and thickening agents in emulsion stability will therefore be another commontheme of this chapter.

7.3. GRAVITATIONAL SEPARATION

In general, the droplets in an emulsion have a different density to that of the liquid whichsurrounds them, and so a net gravitational force acts upon them (Dickinson and Stainsby1982, Hunter 1989, Dickinson 1992, Walstra 1996a,b). If the droplets have a lower densitythan the surrounding liquid, they have a tendency to move upward, which is referred to ascreaming (Figure 7.3). Conversely, if they have a higher density than the surrounding liquid,they tend to move downward, which is referred to as sedimentation. The densities of mostedible oils (in their liquid state) are lower than that of water, and so there is a tendency foroil to accumulate at the top of an emulsion and water at the bottom. Thus, droplets in an oil-in-water emulsion tend to cream, whereas those in a water-in-oil emulsion tend to sediment.

Gravitational separation is usually regarded as having an adverse effect on the quality offood emulsions. A consumer expects to see a product which appears homogeneous, andtherefore the separation of an emulsion into an optically opaque droplet-rich layer and a lessopaque droplet-depleted layer is undesirable. The textural attributes of a product are alsoadversely affected by gravitational separation, because the droplet-rich layer tends to be moreviscous than expected, whereas the droplet-depleted layer tends to be less viscous. The tasteand mouthfeel of a portion of food therefore depend on the location from which it was takenfrom the creamed emulsion. A sample selected from the top of an oil-in-water emulsion thathas undergone creaming will seem too “rich” because of the high fat content, whereas asample selected from the bottom will seem too “watery” because of the low fat content.Gravitational separation is also a problem because it causes droplets to come into close


contact for extended periods, which can lead to enhanced flocculation or coalescence andeventually to oiling off, which is the formation of a layer of pure oil on top of the emulsion.When a food manufacturer is designing an emulsion-based product, it is therefore importantto control the rate at which gravitational separation occurs.

Each food product is unique, containing different types of ingredients and experiencingdifferent environmental conditions during its processing, storage, and consumption. As aconsequence, the optimum method of controlling gravitational separation varies from productto product. In this section, we consider the most important factors which influence gravita-tional separation, as well as strategies for controlling it.

7.3.1. Physical Basis of Gravitational Separation

7.3.1.1. Stokes’ Law

The rate at which an isolated spherical particle creams in an ideal liquid is determined by thebalance of forces which act upon it (Figure 7.3 inset). When a particle has a lower densitythan the surrounding liquid, an upward gravitational force acts upon it (Hiemenz 1986,Hunter 1989):

F r gg = − −43

32 1π ρ ρ( ) (7.7)

where r is the radius of the particle, g is the acceleration due to gravity, ρ is the density, andthe subscripts 1 and 2 refer to the continuous and dispersed phases, respectively. As theparticle moves upward through the surrounding liquid, it experiences a hydrodynamic fric-tional force that acts in the opposite direction and therefore retards its motion:

F rvf = 6 1πη (7.8)

where v is the creaming velocity and η is the shear viscosity. The particle rapidly reaches aconstant velocity, where the upward force due to gravity balances the downward force dueto friction (i.e., Fg = Ff). By combining Equations 7.7 and 7.8, we obtain the Stokes’ lawequation for the creaming rate of an isolated spherical particle in a liquid:

vgr

Stokes = −−2

9

22 1

1

( )ρ ρη (7.9)

The sign of v determines whether the droplet moves upward (+) or downward (–). To a firstapproximation, the stability of a food emulsion to creaming can be estimated using Equation7.9. For example, an oil droplet (ρ2 = 910 kg m–3) with a radius of 1 µm suspended in water(η1 = 1 mPa s, ρ1 = 1000 kg m–3) will cream at a rate of about 17 mm/day. Thus one wouldnot expect an emulsion containing droplets of this size to have a particularly long shelf life.As a useful rule of thumb, an emulsion in which the creaming rate is less than about 1 mm/day can be considered to be stable toward creaming (Dickinson 1992).

In the rest of this section, we shall mainly consider creaming, rather than sedimentation,because it is more common in food systems. Nevertheless, the same physical principles areimportant in both cases, and the methods of controlling them are similar. In the initial stagesof creaming, the droplets move upward and a droplet-depleted layer is observed at thebottom of the container (Figure 7.4). When the droplets reach the top of the emulsion, theycannot move upward any further, and so they pack together to form a “creamed layer.” The


FIGURE 7.4 Time dependence of the creaming process in emulsions. The droplets move upward untilthey cannot move any further and form a “creamed” layer.

thickness of the creamed layer depends on the effectiveness of the droplet packing. Dropletsmay pack tightly or loosely depending on their polydispersity and the nature of the interac-tions between them. Tightly packed droplets tend to form a thin creamed layer, whereasloosely packed droplets form a thick creamed layer. Many of the factors which determinethe packing of droplets in a creamed layer also determine the structure of flocs (see Section7.4). The droplets in a creamed emulsion can often be redispersed by mild agitation,provided they are not too strongly attracted to each other or that coalescence has notoccurred.

7.3.1.2. Modifications of Stokes’ Law

Stokes’ law can only be strictly used to calculate the velocity of an isolated rigid sphericalparticle suspended in an ideal liquid of infinite extent (Hiemenz 1986, Dickinson 1992). Inpractice, there are often large deviations between the creaming velocity predicted by Stokes’law and experimental measurements of creaming in food emulsions, because many of theassumptions used in deriving Equation 7.9 are invalid. Some of the most important factorsthat alter the creaming rate in food emulsions are considered below.

Droplet Fluidity. Stokes’ equation assumes there is no slip at the interface between thedroplet and the surrounding fluid, which is only strictly true for solid particles. The liquidwithin a droplet can move when a force is applied to the surface of the droplet; thus thefrictional force that opposes the movement of a droplet is reduced, which causes an increasein the creaming velocity (Dickinson and Stainsby 1982):

v v=++Stokes

3

3 22 1

2 1

( )

( )

η ηη η (7.10)

This expression reduces to Stokes’ equation when the viscosity of the droplet is muchgreater than that of the continuous phase (η2 >> η1). Conversely, when the viscosity of thedroplet is much less than that of the continuous phase (η2 << η1), the creaming rate is 1.5times faster than that predicted by Equation 7.9. In practice, the droplets in most foodemulsions can be considered to act like rigid spheres because they are surrounded by aviscoelastic interfacial layer which prevents the fluid within them from moving (Walstra1996a,b).

Nondilute Systems. The creaming velocity of droplets in concentrated emulsions is less thanthat in dilute emulsions because of hydrodynamic interactions between the droplets (Hunter


1989, Walstra 1996b). As an emulsion droplet moves upward due to gravity, an equal volumeof continuous phase must move downward to compensate. Thus there is a net flow ofcontinuous phase downward, which opposes the upward movement of the droplets andtherefore decreases the creaming velocity. In fairly dilute emulsions (i.e., <2% droplets), thecreaming velocity of droplets is given by (Hunter 1989):

v v= −Stokes( . )1 6 55φ (7.11)

In more concentrated emulsions, a number of other types of hydrodynamic interaction alsoreduce the creaming velocity of the droplets. These hydrodynamic effects can be partlyaccounted for by using a value for the viscosity of a concentrated emulsion, rather than thatof the continuous phase, in Equation 7.9 (see Section 8.4).

A semiempirical equation that has been found to give relatively good predictions of thecreaming behavior of concentrated emulsions has been developed (Hunter 1989):

v vc

k c

= −

Stokes 1

φφ

φ

(7.12)

Here, φc and k are parameters which depend on the nature of the emulsion. This equationpredicts that the creaming velocity decreases as the droplet concentration increases, untilcreaming is completely inhibited once a critical dispersed-phase volume fraction (φc) hasbeen exceeded (Figure 7.5). In general, the value of φc depends on the packing of the dropletswithin an emulsion, which is governed by their polydispersity and colloidal interactions.Polydisperse droplets are able to fill the available space more effectively than monodispersedroplets because the small droplets can fit into the gaps between the larger ones (Das andGhosh 1990), and so φc is increased. When the droplets are strongly attracted to each other,they can form a particle gel at relatively low droplet concentrations, which prevents anydroplet movement (Figure 7.6). When the droplets are strongly repelled from each other, theireffective size increases, which also causes complete restriction of their movement at lowervalues of φc.

Polydispersity. Food emulsions contain a range of different droplet sizes, and the largerdroplets cream more rapidly than the smaller droplets, so that there is a distribution ofcreaming rates within an emulsion (Figure 7.4). As the larger droplets move upward more

FIGURE 7.5 Prediction of the creaming rate as a function of dispersed-phase volume fraction inconcentrated emulsions (Equation 7.11). Notice that creaming is effectively prevented when the emul-sion droplets become closely packed.


rapidly, they collide with smaller droplets (Melik and Fogler 1988, Dukhin and Sjoblom1996). If the droplets aggregate after a collision, they will cream at a faster rate than eitherof the isolated droplets. Detailed information about the evolution of the droplet concentrationthroughout an emulsion can be obtained by using computer simulations which take intoaccount polydispersity (Davis 1996, Tory 1996). For many purposes, it is only necessary tohave an average creaming velocity, which can be estimated by using a mean droplet radius(r54) in the Stokes equation (Dickinson 1992, Walstra 1996b).

Droplet Flocculation. In many food emulsions, the droplets aggregate to form flocs (Sec-tion 7.4). The size and structure of the flocs within an emulsion have a large influence on therate at which the droplets cream (Bremer 1992, Bremer et al. 1993, Walstra 1996a, Pinfieldet al. 1997). At low or intermediate droplet concentrations, where flocs do not substantiallyinteract with one another, flocculation tends to increase the creaming velocity because theflocs have a larger effective size than the individual droplets (which more than compensatesfor the fact that the density difference between the flocs and the surrounding liquid isreduced). In concentrated emulsions, flocculation retards creaming because a three-dimen-sional network of aggregated flocs is formed, which prevents the individual droplets frommoving (Figure 7.6). The droplet concentration at which creaming is prevented depends onthe structure of the flocs formed. A network can form at lower dispersed-phase volumefractions when the droplets in a floc are more loosely packed, and therefore creaming isprevented at lower droplet concentrations (Figure 7.6). These loosely packed flocs tend toform when there is a strong attraction between the droplets (Section 7.4.3).

Non-Newtonian Rheology of Continuous Phase. The continuous phase of many foodemulsions is non-Newtonian (i.e., the viscosity depends on shear rate or has some elasticcharacteristics) (Chapter 8). As a consequence, it is important to consider which is the mostappropriate viscosity to use in Stokes’ equation (van Vliet and Walstra 1989, Walstra 1996a).Biopolymers, such as modified starches or gums, are often added to oil-in-water emulsionsto increase the viscosity of the aqueous phase (Section 4.6). Many of these biopolymersolutions exhibit shear-thinning behavior; that is, they have a high viscosity at low shear rateswhich decreases dramatically as the shear rate is increased (Chapter 8). This property isimportant because it means that the droplets are prevented from creaming, but that the foodemulsion still flows easily when poured from a container (Dickinson 1992). Creaming usuallyoccurs when an emulsion is at rest, and therefore it is important to know the apparentviscosity that a droplet experiences as it moves through the continuous phase under theseconditions. Typically, the shear rate that a droplet experiences as it creams is between about10–4 to 10–7 s–1 (Dickinson 1992). Solutions of thickening agents have extremely high

FIGURE 7.6 Creaming is prevented at lower dispersed-phase volume fractions when the droplets inthe flocs are more loosely packed.


apparent shear viscosities at these low shear rates, which means that the droplets creamextremely slowly (Walstra 1996a).

Some aqueous biopolymer solutions have a yield stress (τB), below which the solution actslike an elastic solid and above which it acts like a viscous fluid (van Vliet and Walstra 1989).In these systems, droplet creaming is effectively eliminated when the yield stress of thesolution is larger than the stress exerted by a droplet as it moves through the continuousphase, i.e., τB ≥ 2r (ρ1 – ρ2)g (Dickinson 1992). Typically, a value of about 10 mPa is requiredto prevent emulsion droplets of a few micrometers from creaming. Similar behavior isobserved in water-in-oil emulsions that contain a network of aggregated fat crystals which hasa sufficiently high yield stress to prevent the water droplets from sedimenting.

The above discussion highlights the importance of carefully defining the rheologicalproperties of the continuous phase. For this reason, it is good practice to measure theviscosity of the continuous phase over the range of shear rates that an emulsion dropletis likely to experience during processing, storage, and handling, which may be as wideas 10–7 to 103 s–1 (Dickinson 1992).

Electrical Charge. Charged emulsion droplets tend to move more slowly than unchargeddroplets for two reasons (Dickinson and Stainsby 1982, Walstra 1986a). First, repulsiveelectrostatic interactions between similarly charged droplets mean that they cannot get asclose together as uncharged droplets. Thus, as a droplet moves upward, there is a greaterchance that its neighbors will be caught in the downward flow of the continuous phase.Second, the cloud of counterions surrounding a droplet moves less slowly than the dropletitself, which causes an imbalance in the electrical charge which opposes the movement of thedroplet.

Fat Crystallization. Many food emulsions contain a lipid phase that is either partly orwholly crystalline (Mulder and Walstra 1974, Boode 1992, Dickinson and McClements1995). In oil-in-water emulsions, the crystallization of the lipid phase affects the overallcreaming rate because solid fat (ρ ≈ 1200 kg m–3) has a higher density than liquid oil (ρ ≈910 kg m–3). The density of a droplet containing partially crystallized oil is given by ρdroplet

= φSFC ρsolid + (1 – φSFC) ρliquid, where φSFC is the solid fat content At a solid fat content ofabout 30%, an oil droplet has a similar density as water and will therefore neither cream norsediment. At lower solid fat contents, the droplets cream, and at higher solid fat contentsthey sediment. This accounts for the more rapid creaming of milk fat globules at 40°C,where they are completely liquid, compared to at 20°C, where they are partially solid(Mayhill and Newstead 1992).

As mentioned above, crystallization of the fat in a water-in-oil emulsion may lead to theformation of a three-dimensional network of aggregated fat crystals, which prevents the waterdroplets from sedimenting (e.g., in butter and margarine) (Dickinson and Stainsby 1982). Inthese systems, there is a critical solid fat content necessary for the formation of a network,which depends on the morphology of the fat crystals (Walstra 1987). The importance ofnetwork formation is illustrated by the effect of heating on the stability of margarine. Whenmargarine is heated above a temperature where most of the fat crystals melt, the networkbreaks down and the water droplets sediment, leading to the separation of the oil and aqueousphases.

Adsorbed Layer. The presence of a layer of adsorbed emulsifier molecules at the surface ofan emulsion droplet affects the creaming rate in a number of ways. First, it increases theeffective size of the emulsion droplet, and therefore the creaming rate is increased by a factorof (1 + δ/r)2, where δ is the thickness of the adsorbed layer. Typically, the thickness of anadsorbed layer is between about 2 and 10 nm, and therefore this effect is only significant forvery small emulsion droplets (<0.1 µm). Second, the adsorbed layer may alter the effective


density of the dispersed phase, ρ2 (Tan 1990). The effective density of the dispersed phasewhen the droplets are surrounded by an adsorbed layer can be calculated using the followingrelationship, assuming that the thickness of the adsorbed layer is much smaller than the radiusof the droplets (δ << r):

ρρ δ ρ

δ2

3

1 3=

++

droplet layer( / )

( / )

r

r(7.13)

The density of the emulsifier layer is usually greater than that of either the continuous ordispersed phase, and therefore the adsorption of emulsifier increases the effective density ofthe dispersed phase (Tan 1990). The density of large droplets (r >> δ) is approximately thesame as that of the bulk dispersed phase, but that of smaller droplets may be altered signifi-cantly. It is therefore possible to retard creaming by using a surface-active biopolymer whichforms a high-density interfacial layer.

Brownian Motion. Another major limitation of Stokes’ equation is that it ignores the effectsof Brownian motion on the creaming velocity of emulsion droplets (Pinfield et al. 1994,Walstra 1996a). Gravity favors the accumulation of droplets at either the top (creaming) orbottom (sedimentation) of an emulsion. On the other hand, Brownian motion favors therandom distribution of droplets throughout the whole of the emulsion because this maximizesthe configurational entropy of the system. The equilibrium distribution of droplets in anemulsion which is susceptible to both creaming and Brownian motion is given by thefollowing equation (ignoring the finite size of the droplets) (Walstra 1996a):

φ φπ ρ

( ) exphr gh

kT=

−

0

34

3

∆(7.14)

where φ(h) is the concentration of the droplets at a distance h below the top of the emulsion,and φ0 is the concentration of the droplets at the top of the emulsion. If φ(h) = φ0, then thedroplets are evenly dispersed between the two locations (i.e., Brownian motion dominates),but if φ(h) << φ0, the droplets tend to accumulate at the top of the emulsion (i.e., creamingdominates). It has been proposed that if φ(h)/φ0 is less than about 0.02, then the influence ofBrownian motion on the creaming behavior of emulsions is negligible (Walstra 1996a). Thiscondition is met for droplets with radii greater than about 25 nm, which is nearly always thecase in food emulsions.

Complexity of Creaming. The above discussion has highlighted the many factors whichneed to be considered when predicting the rate at which droplets cream in emulsions. Inpractice, it is difficult to simultaneously account for all of these factors in a single analyticalequation. The most comprehensive method of predicting gravitational separation in emul-sions it is to use computer simulations (Pinfield et al. 1994, Tory 1996).

7.3.2. Methods of Controlling Gravitational Separation

The discussion of the physical basis of gravitational separation in the previous section hashighlighted a number of ways of retarding its progress in food emulsions.

7.3.2.1. Minimize Density Difference

The driving force for gravitational separation is the density difference between the dropletsand the surrounding liquid: ∆ρ = (ρ2 – ρ1). It is therefore possible to prevent gravitational


separation by “matching” the densities of the oil and aqueous phases (Tan 1990). Mostnaturally occurring edible oils have fairly similar densities (≈910 kg m–3), and therefore foodmanufacturers have limited flexibility in preventing creaming by changing the type of oilused in their products. Nevertheless, a number of alternative strategies have been developedwhich enable food manufacturers to match the densities of the dispersed and continuousphases more closely.

Density matching can be achieved by mixing natural oils with brominated vegetable oils(which have a higher density than water), so that the overall density of the oil droplets issimilar to that of the aqueous phase (Tan 1990). Nevertheless, the utilization of brominatedoils in foods is restricted in many countries, and so food manufacturers only have a limitedamount of control over creaming stability using this method. The restrictions on the use ofbrominated oils has led to the utilization of other types of “weighting agents” which aresoluble in the oil phase and increase the density of the oil droplets, such as ester gum, damargum, and sucrose acetate isobutyrate (Tan 1990). These weighting agents are usually incor-porated into the oil phase prior to the homogenization process.

If the droplets in an oil-in-water emulsion are small, it may be possible to prevent gravi-tational separation by using an emulsifier which forms a relatively thick and dense interfaciallayer, because this decreases the density difference between the oil droplets and the surround-ing liquid (Section 7.3.1).

In some emulsions, it is possible to control the degree of gravitational separation byvarying the solid fat content of the lipid phase. As mentioned in Section 7.3.1, an oil dropletwith a solid fat content of about 30% has a density similar to water and will therefore bestable to gravitational separation. The solid fat content of a droplet could be controlled byaltering the composition of the lipid phase or by controlling the temperature (Dickinson andMcClements 1995). In practice, this procedure is unsuitable for many food emulsions becausepartially crystalline droplets are susceptible to partial coalescence, which severely reducestheir stability (Section 7.6).

7.3.2.2. Reduce Droplet Size

Stokes’ law indicates that the velocity at which a droplet moves is proportional to the squareof its radius (Equation 7.9). The stability of an emulsion to gravitational separation cantherefore be enhanced by reducing the size of the droplets it contains. Homogenization of rawmilk is one of the most familiar examples of the retardation of creaming in a food emulsionby droplet size reduction (Swaisgood 1996). A food manufacturer generally aims to reducethe size of the droplets in an emulsion below some critical radius that is known to be smallenough to prevent them from creaming during the lifetime of the product. In practice,homogenization leads to the formation of emulsions that contain a range of different sizes,and a certain percentage of the droplets is likely to be above the critical radius and thereforesusceptible to gravitational separation. For this reason, a food manufacturer usually specifiesthe minimum percentage of droplets which can be above the critical radius. For example,cream liqueurs are usually designed so that less than 3% of the droplets have diametersgreater than 0.4 µm (Dickinson 1992). Even though a small fraction of the droplets are greaterthan this size, and therefore susceptible to creaming, this does not cause a major problembecause the presence of the droplet-rich layer is obscured by the turbidity produced by thedroplets which remain in the droplet-depleted layer.

7.3.2.3. Modify Rheology of Continuous Phase

Increasing the viscosity of the liquid surrounding a droplet (η1) decreases the velocity atwhich the droplet moves (Equation 7.9). Thus the stability of an emulsion to gravitational


separation can be enhanced by increasing the viscosity of the continuous phase (e.g., byadding a thickening agent) (Section 4.6). Gravitational separation may be completely retardedif the continuous phase contains a three-dimensional network of aggregated molecules orparticles which traps the droplets and prevents them from moving. Thus the droplets in oil-in-water emulsions can be completely stabilized against creaming by using biopolymerswhich form a gel in the aqueous phase, while the droplets in water-in-oil emulsions can becompletely stabilized against sedimentation by ensuring there is a network of aggregated fatcrystals in the oil phase (van Vliet and Walstra 1989).

7.3.2.4. Increase Droplet Concentration

The rate of gravitational separation can be retarded by increasing the droplet concentration.At a sufficiently high dispersed-phase volume fraction, the droplets are prevented frommoving because they are so closely packed together (Figure 7.6). It is for this reason that thedroplets in mayonnaise, which has a high dispersed-phase volume fraction, are more stableto creaming than those in salad dressings, which have a lower dispersed-phase volumefraction. Nevertheless, it should be mentioned that it is often not practically feasible to alterthe droplet concentration, and therefore one of the alternative methods of preventing cream-ing should be used.

7.3.2.5. Alter Degree of Droplet Flocculation

The rate of gravitational separation can be controlled by altering the degree of flocculationof the droplets in an emulsion. In dilute emulsions, flocculation causes enhanced gravitationalseparation because it increases the effective size of the particles. To improve the stability ofthese systems, it is important to ensure that the droplets are prevented from flocculating(Section 7.4). In concentrated emulsions, flocculation reduces the rate of gravitational sepa-ration because the droplets are prevented from moving past one another (Figure 7.6). Thecritical dispersed-phase volume fraction at which separation is prevented depends on thestructural organization of the droplets within the flocs (Section 7.4.3). The stability ofconcentrated emulsions may therefore be enhanced by altering the nature of the colloidalinteractions between the droplets and therefore the structure of the flocs formed.

7.3.3. Experimental Measurement of Gravitational Separation

To theoretically predict the rate at which gravitational separation occurs in an emulsion, it isnecessary to have information about the densities of the dispersed and continuous phases, thedroplet size distribution, and the rheological properties of the continuous phase. The densityof the liquids can be measured using a variety of techniques, including density bottles,hydrometers, and oscillating U-tube density meters (Pomeranz and Meloan 1994). The drop-let size distribution can be measured by microscopy, light-scattering, electrical pulse count-ing, or ultrasonic methods (see Chapter 10). The rheological properties of the continuousphase can be characterized using various types of viscometers and dynamic shear rheometers(see Chapter 8). In principle, it is possible to predict the long-term stability of a food emulsionfrom a knowledge of these physicochemical properties and a suitable mathematical model.In practice, this approach has limited use because the mathematical models are not currentlysophisticated enough to take into account the inherent complexity of food emulsions. For thisreason, it is often more appropriate to directly measure the gravitational separation of thedroplets in an emulsion.

The simplest method of monitoring gravitational separation is to place an emulsion in atransparent test tube, leave it for a certain length of time, and then measure the height of theinterface between the droplet-depleted and droplet-rich layers using a ruler. This procedure


can often be accelerated by centrifuging an emulsion at a fixed speed for a certain length oftime (Sherman 1995). Nevertheless, this accelerated creaming test should be used withcaution because the factors which determine droplet movement in a gravitational field areoften different from those which are important in a centrifugal field. For example, thecontinuous phase may have a yield stress which is exceeded in a centrifuge but which wouldnever be exceeded under normal storage conditions. The two major problems associated withvisually determining the extent of creaming are (1) it is only possible to obtain informationabout the location of the boundary between the droplet-rich layer and droplet-depleted layers,rather than about the full vertical concentration profile of the droplets, and (2) in somesystems, the droplet-depleted layer may be optically opaque, and so it is difficult to clearlylocate the interface between it and the droplet-rich layer.

A more sophisticated method of monitoring gravitational separation is to use light scatter-ing (Davis 1996). An emulsion is placed in a vertical glass tube and a monochromatic beamof near-infrared light is directed through it (Figure 7.7). The percentage of transmitted andscattered light is measured as a function of emulsion height by scanning the light beam upand down the sample using a stepper motor. The variation in droplet concentration withemulsion height can then be deduced from the percentage of transmitted and scattered lightusing a suitable theory or calibration curve. The technique could be used to measure both thesize and concentration of the droplets at any height by measuring the angular dependence ofthe intensity of the scattered light. This device is mainly used to study creaming and sedimen-tation kinetics in fairly dilute emulsions, because it is not possible to transmit light throughconcentrated emulsions.

Traditionally, the kinetics of gravitational separation was monitored in concentrated emul-sions by physically removing sections of an emulsion from different heights and then ana-lyzing the concentration of droplets in each section (e.g., by measuring the density or byevaporating the water) (Pal 1994). These techniques cause the destruction of the sample beinganalyzed and therefore cannot be used to monitor creaming in the same sample as a functionof time. Instead, a large number of similar samples have to be prepared and each one analyzedat a different time. Recently, a number of analytical methods have been developed to monitorgravitational separation in concentrated emulsions without disturbing the sample (e.g., ultra-sonics, dielectric spectroscopy, and nuclear magnetic resonance) (Chapter 10). The ultrasonicdevice is very similar to the light-scattering technique described above, except that it is based

FIGURE 7.7 Light-scattering device for vertically scanning the creaming and sedimentation of drop-lets in emulsions.


on the propagation of ultrasonic waves through an emulsion rather than electromagneticwaves. An ultrasonic transducer is scanned vertically up and down an emulsion, whichenables one to determine both the size and concentration of the droplets as a function ofemulsion height (Chapter 10). Nuclear magnetic resonance imaging techniques, which arebased on differences in the response of oil and water to the application of a radio-frequencypulse, have also been used to monitor gravitational separation in emulsions (Chapter 10).These techniques enable one to obtain a three-dimensional image of the droplet concentrationwithin an emulsion, but the equipment is expensive to purchase and requires highly skilledoperators, which has limited their application (Dickinson and McClements 1995, Sodermanand Bailnov 1996).

7.4. FLOCCULATION

The droplets in emulsions are in continual motion because of the effects of thermal energy,gravity, or applied mechanical forces, and as they move about, they frequently collide withtheir neighbors (Lips et al. 1993, Dukhin and Sjoblom 1996). After a collision, emulsiondroplets may either move apart or remain aggregated, depending on the relative magnitudeof the attractive and repulsive interactions between them (Chapter 3). Droplets aggregatewhen there is a minimum in the interdroplet pair potential which is sufficiently deep andwhich is accessible to the droplets. The two major types of aggregation in food emulsions areflocculation and coalescence (Dickinson and Stainsby 1982; Dickinson 1992; Walstra 1996a,b).Flocculation is the process whereby two or more droplets come together to form an aggregatein which the droplets retain their individual integrity, whereas coalescence is the processwhereby two or more droplets merge together to form a single larger droplet. Dropletflocculation is discussed in this section, and droplet coalescence is covered in Section 7.5.

Droplet flocculation may be either advantageous or detrimental to emulsion quality de-pending on the nature of the food product. Flocculation accelerates the rate of gravitationalseparation in dilute emulsions, which is undesirable because it reduces their shelf life (Luytenet al. 1993). It also causes a pronounced increase in emulsion viscosity and may even leadto the formation of a gel (Demetriades et al. 1997a,b). Some food products are expected tohave a low viscosity, and therefore flocculation is detrimental. In other products, a controlledamount of flocculation may be advantageous because it leads to the creation of a desirabletexture. Improvements in the quality of emulsion-based food products therefore depend on abetter understanding of the factors which determine the degree of floc formation, the structureof the flocs formed, and the rate at which flocculation proceeds. In addition, it is importantto understand the effect that flocculation has on the bulk physicochemical properties ofemulsions.

7.4.1. Physical Basis of Flocculation

As flocculation proceeds, there is a decrease in the total number of particles (monomers +aggregates) in the emulsion, which can be described by the following equation:

dn

dtFET = − 1

2 (7.15)

where dnT/dt is the flocculation rate, nT is the total number of particles per unit volume, t isthe time, F is the collision frequency, and E is the collision efficiency. A factor of one-halfappears in the equation because a collision between two particles leads to a reduction of onein the total number of particles present. Equation 7.15 indicates that the rate at which


flocculation occurs depends on two factors: the frequency of collisions between the dropletsand the fraction of collisions which leads to aggregation.

7.4.1.1. Collision Frequency

The collision frequency is the total number of droplet encounters per unit time per unitvolume of emulsion. Any factor which increases the collision frequency increases the floc-culation rate (provided that it does not also decrease the collision efficiency). Collisionsbetween droplets occur as a result of their movement, which may be induced by Brownianmotion, gravitational separation, or applied mechanical forces.

Collisions Due to Brownian Motion. In quiescent systems, the collisions between dropletsare mainly a result of their Brownian motion. By considering the diffusion of particles in adilute suspension, von Smoluchowski was able to derive the following expression for thecollision frequency (Dickinson and Stainsby 1982):

F D rnB = 16 02π (7.16)

Here, FB is the collision frequency due to Brownian motion (m–3 s–1), D0 is the diffusioncoefficient of a single particle (m2 s–1), n is the number of particles per unit volume (m–3),and r is the droplet radius (m). For spherical particles, D0 = kT/6πη1r, where η1 is theviscosity of the continuous phase, k is Boltzmann’s constant, and T is the absolute tempera-ture. Hence:

F k nkTn kT

rB B= = =22

1

2

16

83

3

2ηφ

η π (7.17)

where kB is a second-order rate constant (m3 s–1) and φ is the dispersed-phase volume fraction.For particles dispersed in water at room temperature, the collision frequency is ≈2 × 1018 φ2/r 6, when the radius is expressed in micrometers. Equation 7.17 indicates that the frequencyof collisions between droplets can be reduced by decreasing their volume fraction, increasingtheir size, or increasing the viscosity of the continuous phase. If it is assumed that everycollision between two particles leads to aggregation, and that the rate constant is independentof aggregate size, then the flocculation rate is given by dnT/dt = –1

2 FB, which can be integratedto give the following expression for the change in the total number of particles with time(Evans and Wennerstrom 1994):

nn

k n tTB

=+

0

01 12

(7.18)

where n0 is the initial number of particles per unit volume. The time taken to reduce thenumber of droplets in an emulsion by half can be calculated from the above equation:

τη πη

φ1 20

1

0

13

0

2 3

4/ = = =

k n kTn kT

r

B(7.19)

For a system where the particles are suspended in water at room temperature, τ1/2 ≈ r 3/φ0,when r is expressed in micrometers. Thus, an oil-in-water emulsion with φ = 0.1 and r = 1µm would have a half-life of about 10 s, which is on the same order as the existence of an


FIGURE 7.8 Dependence of the concentration of the total number of particles (nT), monomers (n1),dimers (n2), and trimers (n3) on time t /τ1/2. The number of monomers decreases with time, whereas thenumber of aggregates initially increases and then decreases.

emulsion prepared by shaking oil and water together in the absence of a stabilizer oremulsifier. It is also possible to derive an equation to describe the change in the number ofdimers, trimers, and other aggregates with time (Evans and Wennerstrom 1994):

n nt t

k

k k

=

+

− − −

01 2

1

1 2

1

1τ τ/ /

(7.20)

where nk is the number of aggregates per unit volume containing k particles. The predictedvariation in the total concentration of particles and of the concentration of monomers (k = 1),dimers (k = 2), and trimers (k = 3) with time is shown in Figure 7.8. As would be expected,the total number of particles and the number of monomers decrease progressively with timeas flocculation proceeds. The number of dimers, trimers, and other aggregates initiallyincreases with time and then decreases as they interact with other particles and form largeraggregates.

The above equations are only applicable to dilute suspensions containing identical spheri-cal particles suspended in an ideal liquid. Many of the assumptions used in their derivationare not valid for actual food emulsions, which may be concentrated, polydisperse, and havenonideal continuous phases. In addition, the properties of the flocs cannot be assumed to bethe same as those of the monomers, and therefore the above theory has to be modified to takeinto account the dimensions, structure, and hydrodynamic behavior of the flocs (Bremer1992, Walstra 1996a).

Collisions Due to Gravitational Separation. In polydisperse emulsions, droplet–dropletencounters can occur because of the different creaming (or sedimentation) rates of thedifferently sized droplets. Large droplets move more quickly than smaller ones and thereforecollide with them as they move upward (or downward). The collision frequency for gravita-tionally induced flocculation is given by (Melik and Fogler 1988, Zhang and Davis 1991):

F v v r r n nG = − +π( )( )2 1 1 22

1 2 (7.21)

F k n ng r r r r

r rG G= =− +

1 2

1 2

1

22

12

1 22

13

238

∆ρφ φπη

( )( )(7.22)


where FG is the collision frequency due to gravitational separation, vi is the Stokes creamingvelocity of a particle with radius ri , and ∆ρ is the density difference between the droplets andthe surrounding liquid. This equation indicates that the collision frequency increases as thedifference between the creaming velocities of the particles increases. The rate of gravitation-ally induced flocculation can therefore be retarded by ensuring that the droplet size distribu-tion is not too wide, decreasing the density difference between the oil and aqueous phases,decreasing the droplet concentration, or increasing the viscosity of the continuous phase.Equation 7.22 would have to be modified before it could be applied to systems which do notobey Stokes’ law (Section 7.3.1). In addition, it does not take into account the fact that thedroplets reach a position at the top or bottom of an emulsion where they cannot move anyfurther and are therefore forced to encounter each other.

Collisions Due to Applied Shear Forces. Food emulsions are often subjected to variouskinds of shear flow during their production, storage, and transport. Consequently, it isimportant to appreciate the effect that shearing has on their stability to flocculation. In asystem subjected to Couette flow, the collision frequency is given by (Dickinson 1992,Walstra 1996a):

F k n Gr nG

rS S= = =

2 3 22

3

163

3π

φ(7.23)

where FS is the collision frequency due to shear. Thus the frequency of shear-inducedcollisions can be retarded by decreasing the shear rate, increasing the droplet size, or decreas-ing the dispersed-phase volume fraction. It should be noted that the collision frequency isindependent of the viscosity of the continuous phase.

Relative Importance of Different Collision Mechanisms. In general, each of the abovemechanisms may contribute to the droplet collision frequency in an emulsion. In practice, oneof the mechanisms usually dominates, depending on the composition and microstructure ofthe product and the prevailing environmental conditions. To effectively control the collisionfrequency, it is necessary to establish which mechanism is the most important in the particularsystem being studied. The ratio of the shear-to-Brownian motion collision frequencies (FS/FB)and the gravitational-to-Brownian motion collision frequencies (FG/FB) is plotted as a func-tion of shear rate and particle size ratio (= r2 /r1) in Figure 7.9 for a typical emulsion. At lowshear rates (G < 2 s–1), collisions due to Brownian motion dominate, but at high shear ratesthose due to mechanical agitation of the system dominate. Gravitationally induced collisionsdominate those due to Brownian motion when the particle size ratio exceeds about 5, and thusit is likely to be important in emulsions which have a broad particle size distribution.

7.4.1.2. Collision Efficiency

If every encounter between two droplets led to aggregation, then emulsions would not remainstable long enough to be practically useful. To prevent droplets from flocculating during acollision, it is necessary to have a sufficiently high repulsive energy barrier to stop them fromcoming too close together (Chapter 3). The height of this energy barrier determines thelikelihood that a collision will lead to flocculation (i.e., the collision efficiency). The collisionefficiency (E) has a value between 0 (no flocculation) to 1 (every collision leads to floccu-lation)* and depends on the hydrodynamic and colloidal interactions between the droplets.The flocculation rate therefore depends on the precise nature of the interactions between the

* In practice, E can have a value which is slightly higher than 1 because droplet collisions are accelerated whenthere is a strong attraction between the droplets.


(A) (B)

emulsion droplets. For collisions induced by Brownian motion (Dukhin and Sjoblom 1996,Walstra 1996b):

− =dn

dt

kTn EB B4

3

2

1η (7.24)

Ew s kT

s G sdsB =

∞ −

∫22

2

1exp[ ( ) / ]

( ) (7.25)

where s is the dimensionless center-to-center distance between the droplets (s = [2r + h]/r),r is the droplet radius, and h is the surface-to-surface separation. Colloidal interactions areaccounted for by the w(s) term and hydrodynamic interactions by the G(s) term (Chapter 3).When there are no colloidal interactions between the droplets, w(s) = 0, and no hydrodynamicinteractions, G(s) = 0, Equation 7.24 becomes equivalent to that derived by Smoluchowski(Equations 7.15 and 7.16). The stability of an emulsion to aggregation is governed primarilyby the maximum height of the energy barrier, wmax(s), rather than by its width (Friberg 1997).To enhance the stability of an emulsion against flocculation, it is necessary to have an energybarrier which is large enough to prevent the droplets from coming close together. An energybarrier of about 20 kT is usually sufficient to provide long-term stability of emulsions (Friberg1997). Expressions for the efficiency of shear and gravitationally induced collisions alsodepend on colloidal and hydrodynamic interactions and have been derived for some simplesystems (Zhang and Davis 1991).

7.4.2. Methods of Controlling Flocculation

A knowledge of the physical basis of droplet flocculation facilitates the development ofeffective strategies for controlling it in food emulsions. These strategies can be convenientlydivided into those which influence the collision frequency and those which influence thecollision efficiency.

FIGURE 7.9 Relative importance of the different collision mechanisms for typical food emulsions(∆ρ = 90 kg m–3, r = 1 µm, φ = 0.1). (A) Shear-induced collisions become increasingly important asthe shear rate is increased. (B) Gravitationally induced collisions become increasingly important as theratio of droplet size increases.


7.4.2.1. Collision Frequency

The rate at which flocculation proceeds can be controlled by manipulating the dropletcollision frequency. The most effective means of achieving this depends on the dominantcollision mechanism in the emulsion (i.e., Brownian motion, gravity, or mechanical agita-tion). The rate at which droplets encounter each other in an unstirred emulsion can bereduced by increasing the viscosity of the continuous phase (Equations 7.17 and 7.22).Flocculation may be completely retarded if the continuous phase contains a three-dimen-sional network of aggregated molecules or particles that prevents the droplets from moving(e.g., a biopolymer gel or a fat crystalline network). The collision frequency increases whenan emulsion is subjected to sufficiently high shear rates (Equation 7.23), and therefore itmay be important to ensure that a product is protected from mechanical agitation during itsstorage and transport in order to avoid flocculation. The collision frequency increases as thedroplet concentration increases or the droplet size decreases, with the precise nature of thisdependence determined by the type of collision mechanism which dominates. The rate ofcollisions due to gravitational separation depends on the relative velocities of the particlesin an emulsion and therefore decreases as the density difference between the droplet andsurrounding liquid decreases or as the viscosity of the continuous phase increases (Equation7.22).

7.4.2.2. Collision Efficiency

The most effective means of controlling the rate and extent of flocculation in an emulsion isto regulate the colloidal interactions between the droplets. Flocculation can be prevented bydesigning an emulsion in which the repulsive interactions between the droplets are signifi-cantly greater than the attractive interactions. A wide variety of different types of colloidalinteraction can act between the droplets in an emulsion (Chapter 3). Which of these isimportant in a given system depends on the type of ingredients present, the microstructure ofthe emulsion, and the prevailing environmental conditions. To control flocculation in aparticular system, it is necessary to identify the most important types of colloidal interaction.

Electrostatic Interactions. Many oil-in-water emulsions used in the food industry are sta-bilized against flocculation by using electrically charged emulsifiers which generate anelectrostatic repulsion between the droplets (e.g., ionic surfactants, proteins, or polysaccha-rides) (Dickinson 1992, Das and Kinsella 1990). The flocculation stability of electrostaticallystabilized emulsions depends mainly on the electrical properties of the emulsifier and the pHand ionic strength of the aqueous phase (Hunter 1986). The number, position, sign, anddissociation constants of the ionizable groups on an emulsifier determine its electrical behav-ior under different environmental conditions. For each type of food product, it is thereforenecessary to select an emulsifier with appropriate electrical characteristics.

Hydrogen ions are potential determining ions for many food emulsifiers (e.g., COOH →COO– + H+ or NH2 + H+ → NH3

+), and therefore the sign and magnitude of the electricalcharge on emulsion droplets are determined principally by the pH of the surrounding solution(Section 3.4.1). The influence of pH on droplet flocculation in a protein-stabilized emulsionis illustrated in Figure 7.10. At pH values sufficiently above or below the isoelectric point(IEP) of the whey proteins (IEP ≈ pH 5), the droplet charge is large enough to preventflocculation because of the strong electrostatic repulsion between the droplets (Figure 7.10).At pH values near the isoelectric point (IEP ± 1), the net charge on the proteins is relativelylow and the electrostatic repulsion between the droplets is no longer sufficiently strong toprevent flocculation. Droplet flocculation leads to a pronounced increase in the viscosity ofan emulsion, as well as a decrease in creaming stability, and therefore has important impli-cations for food quality (Demetriades et al. 1997a, Agboola and Dalgleish 1996d).


The ionic strength of an aqueous solution depends on the concentration and valency of theions it contains (Israelachvili 1992). As the ionic strength is increased, the electrostaticrepulsion between droplets is progressively screened, until eventually it is no longer strongenough to prevent flocculation (Section 3.11.1). The minimum amount of electrolyte re-quired to cause flocculation is known as the critical flocculation concentration or CFC. TheCFC decreases as the surface potential of the emulsion droplets decreases and as the valencyof the counterions increases. It has been shown that CFC ∝ Ψ0

4/z2 (where Ψ0 is the surfacepotential and z is the counterion valency) for droplets with relatively low surface potentials(i.e., Ψ0 < 25 mV) (Hunter 1986). These low surface potentials are found in many foodemulsions which are susceptible to flocculation. Under certain conditions, Ψ0 is inverselyproportional to the valency of the counterions, so that CFC ∝ 1/z6, which is known as theSchultz–Hardy rule (Hunter 1986). This relationship indicates that a much lower concentra-tion of a multivalent ion than a monovalent ion is required to cause flocculation. TheSchultz–Hardy rule can be derived from the DLVO theory by assuming that the CFC occurswhen the potential energy barrier, which normally prevents droplets from aggregating, fallsto a value of zero due to the addition of salt. Consequently, when two droplets collide witheach other, they immediately aggregate into the primary minima. In practice, significantdroplet flocculation occurs when the potential energy barrier is slightly higher than zero, andtherefore the Schultz–Hardy rule is expected to slightly overestimate the CFC (Hunter1986).

Some types of ionic species are capable of specifically binding to oppositely chargedemulsifier groups and can therefore alter the surface charge. Specifically adsorbed ions mayeither decrease, increase, or reverse the charge on a droplet depending on their valency andthe nature of the specific interactions involved (Section 3.4.2.1). In addition to their influenceon surface charge, specifically adsorbed ions are often highly hydrated and therefore increasethe short-range hydration repulsion between droplets (Section 3.8).

The presence of multivalent ions in the continuous phase of an electrostatically stabilizedemulsion has a strong influence on its flocculation stability (Dickinson et al. 1992; Agboolaand Dalgleish 1995, 1996a), first because multivalent ions are much more effective atscreening the electrostatic interactions between droplets than monovalent ions and, second,because they are capable of forming electrostatic bridges between two droplets which havethe same charge (Section 3.4.4). For example, the addition of calcium ions (Ca2+) to a protein-stabilized emulsion that contains negatively charged droplets can cause droplet flocculationbecause the ions shield the electrostatic repulsion between the droplets and because they formprotein–calcium–protein cross-links between the droplets (Dickinson et al. 1992; Agboolaand Dalgleish 1995, 1996a). These multivalent ions may be simple ions (e.g., Ca2+, Mg2+, or

FIGURE 7.10 Influence of pH and ionic strength on the stability of 20% corn-oil-in-water emulsionsto flocculation. Extensive flocculation is observed near the isoelectric point of the proteins.


Al3+), small organic ions (amino acids), or charged biopolymers (e.g., proteins or polysaccha-rides). The influence of multivalent ions on emulsion stability can be reduced by ensuring thatthey are excluded from the formulation or by adding ingredients which form complexes withthem (e.g., EDTA in the case of calcium ions).

The flocculation stability of emulsions that contain electrically charged droplets can becontrolled in a variety of ways depending on the system. To prevent flocculation, it isnecessary to ensure that the droplets have a sufficiently high surface potential under theprevailing environmental conditions (which often requires that the pH be controlled) andthat the electrolyte concentration is below the CFC. In some important types of food, itis necessary to have relatively high concentrations of electrolyte present for nutritionalpurposes (e.g., mineral-fortified emulsions used in infant, elderly, and athlete formulations).In these systems, the influence of the electrolyte on flocculation stability can be reducedby changing to a nonionic emulsifier or by incorporating ingredients which complex themineral.

Polymeric Steric Interactions. Many food emulsifiers prevent droplet flocculation throughpolymeric steric repulsion (Section 3.5). This repulsion must be sufficiently strong and longrange to overcome any attractive interactions (Section 3.11.3). Sterically stabilized emulsionsare much less sensitive to variations in pH and ionic strength than electrostatically stabilizedemulsions (Hunter 1986). Nevertheless, they can become unstable to flocculation undercertain conditions. If the composition of the continuous phase or the temperature is alteredso that polymer–polymer interactions become more favorable than solvent–solvent/solvent–polymer interactions, then the mixing contribution to the steric interaction becomes attractiveand may lead to droplet flocculation. A sterically stabilized emulsion may also becomeunstable if the thickness of the interfacial membrane is reduced (Section 3.11.3), which couldoccur if the polymeric segments on the emulsifier were chemically or biochemically cleaved(e.g., by acid or enzyme hydrolysis) or if the continuous phase becomes a poor solvent forthe polymer segments. Short-range hydration forces make an important contribution to theflocculation stability of many sterically stabilized emulsions (Israelachvili 1992, Evans andWennerstrom 1994). In these systems, droplet flocculation may occur when the emulsion isheated, because emulsifier head groups are progressively dehydrated with increasing tem-perature (Israelachvili 1992, Aveyard et al. 1990).

Biopolymer Bridging. Many types of biopolymer promote flocculation by forming bridgesbetween two or more droplets (Lips et al. 1991). Biopolymers may adsorb either directly tothe bare surfaces of the droplets or to the adsorbed emulsifier molecules that form theinterfacial membrane (Walstra 1996a). To be able to bind to the droplets, there must be asufficiently strong attractive interaction between segments of the biopolymer and the dropletsurface. The most common types of interaction that operate in food emulsions are hydropho-bic and electrostatic (Dickinson 1989, 1992).

When a biopolymer has a number of nonpolar residues along its backbone, some of themmay associate with hydrophobic patches on one droplet, while others associate with hydro-phobic patches on another droplet. This type of bridging flocculation tends to occur when abiopolymer is used as an emulsifier and there is insufficient present to completely cover theoil–water interface formed during homogenization (Walstra 1996a). Bridging may occureither during the homogenization process or after it is complete (e.g., when a biopolymer isonly weakly associated with a droplet, some of its segments can desorb and become attachedto a neighboring droplet). This type of bridging flocculation can usually be prevented byensuring there is a sufficiently high concentration of biopolymer present in the continuousphase prior to homogenization (Dickinson and Euston 1991, Dickinson 1992, Stoll and Buffle1996).


FIGURE 7.11 Influence of temperature on the flocculation stability of 20% corn-oil-in-water emul-sions stabilized by whey protein isolate (pH 7, 0 mM NaCl). Flocculation is observed when theemulsions are heated above 70°C because of protein unfolding and exposure of hydrophobic groups.

Bridging flocculation can also occur when a biopolymer in the continuous phase has anelectrical charge which is opposite to that of the droplets (Pal 1996). In this case, bridgingflocculation can be avoided by ensuring the droplets and biopolymer have similar charges orthat either the droplets or biopolymer are uncharged.

Hydrophobic Interactions. This type of interaction is important in emulsions that containdroplets which have nonpolar regions exposed to the aqueous phase. Their role in influencingthe stability of food emulsions has largely been ignored, probably because of the lack oftheories to describe them and experimental techniques to quantify them. One of the clearestexamples of their importance in food emulsions is the effect of heat on the flocculationstability of oil-in-water emulsions stabilized by globular proteins (Hunt and Dalgleish 1995,Demetriades et al. 1997b). At room temperature, whey protein stabilized emulsions (pH 7,0 mM NaCl) are stable to flocculation because of the large electrostatic repulsion between thedroplets, but when they are heated above 70°C, they become unstable (Figure 7.11). Theglobular proteins adsorbed to the surface of the droplets unfold above this temperature andexpose nonpolar amino acids which were originally located in their interior (Monahan et al.1996, Dalgleish 1996a). Exposure of these nonpolar amino acids increases the hydrophobiccharacter of the droplet surface and therefore leads to flocculation because of the increasedhydrophobic attraction between the droplets (Hunt and Dalgleish 1995, Monahan et al. 1996,Demetriades et al. 1997b).

Hydrophobic interactions are also likely to be important in emulsions in which there is notenough emulsifier present to completely saturate the surfaces of the droplets. This may occurwhen there is insufficient emulsifier present in an emulsion prior to homogenization or whenan emulsion is so diluted that some of the emulsifier desorbs from the droplet surfaces.

Flocculation due to hydrophobic interactions can be avoided by ensuring that there issufficient emulsifier present to completely cover the droplet surfaces or by selecting anemulsifier which does not undergo detrimental conformational changes at the temperaturesused during processing, storage, or handling.

Depletion Interactions. The presence of nonadsorbing colloidal particles, such as biopoly-mers or surfactant micelles, in the continuous phase of an emulsion causes an increase in theattractive force between the droplets due to an osmotic effect associated with the exclusionof colloidal particles from a narrow region surrounding each droplet (Section 3.6). Thisattractive force increases as the concentration of colloidal particles increases, until eventuallyit may become large enough to overcome the repulsive interactions between the droplets andcause them to flocculate (Aronson 1991; Jenkins and Snowden 1996; Dickinson and Golding


1997a,b; Dickinson et al. 1997). This type of droplet aggregation is usually referred to asdepletion flocculation (Walstra 1996a,b). The lowest concentration required to cause deple-tion flocculation is referred to as the critical flocculation concentration by analogy to theCFC used to characterize the effect of salt on the stability of electrostatically stabilizedemulsions. The CFC decreases as the size of the emulsion droplets increases (Section 3.6.3).The flocculation rate initially increases as the concentration of nonadsorbing colloidal par-ticles is increased because of the enhanced attraction between the droplets (i.e., a highercollision efficiency). However, once the concentration of colloidal particles exceeds a certainconcentration, the flocculation rate may actually decrease because the viscosity of the con-tinuous phase increases so much that the movement of the droplets is severely retarded (i.e.,a lower collision frequency). This is clearly illustrated in Figure 7.12, which shows thedependence of the height of the creamed layer in a series of corn-oil-in-water emulsionscontaining different concentrations of xanthan (Basaran et al. 1998). Xanthan is a nonadsorbingbiopolymer which is normally used as a stabilizer or thickening agent in food emulsions(BeMiller and Whistler 1996). In the absence of xanthan, the emulsions are stable to creamingover a 24-h period. At xanthan concentrations around 0.01%, there is a net attraction betweenthe droplets which causes them to flocculate and therefore cream rapidly. At higher xanthanconcentrations, there is still a strong attraction between the droplets, but they are unable tomove because of the large increase in the viscosity of the continuous phase.* Similar obser-vations have been obtained for emulsions to which ionic or nonionic surfactant micelles havebeen added (Bibette 1991, Aronson 1992, McClements 1994, Jenkins and Snowden 1996).For ionic colloidal particles, such as sodium dodecyl sulfate (SDS) micelles or chargedbiopolymers, the CFC is expected to be strongly dependent on electrolyte concentrationbecause of its ability to reduce the electrostatic repulsion between colloidal particles andtherefore reduce their effective size.

Hydrodynamic Interactions. The efficiency of the collisions between droplets is also deter-mined by the strength of the hydrodynamic interactions between them (Davis et al. 1989,Dukhin and Sjoblom 1996). As two droplets approach each other, a repulsion arises because

* Even when the droplets do flocculate at higher xanthan concentrations, the movement of the flocs themselves willbe retarded, and therefore no creaming is observed.

FIGURE 7.12 Influence of xanthan concentration on the stability of 20% corn-oil-in-water emulsionsto depletion flocculation. Flocculation is observed above the critical flocculation concentration (i.e., %xanthan > 0.0075 wt%).


of the resistance associated with the flow of the continuous phase from the thin gap betweenthem. The magnitude of this resistance decreases as the droplet surfaces become more mobile,leading to an increase in the collision efficiency (Section 3.10). On the other hand, thecollision efficiency may be reduced when the droplet surfaces are stabilized by small-molecule surfactants because of the Gibbs–Marangoni effect (Walstra 1996a,b).

Influence of Droplet Size. It should be noted that the magnitude of the colloidal interactionsbetween emulsion droplets usually increases with droplet size, which will cause an increasein the height of any energy barriers and in the depth of any energy minima. As a consequence,altering the droplet size may either increase or decrease the stability of an emulsion, depend-ing on the nature of the system.

7.4.3. Structure and Properties of Flocculated Emulsions

The appearance, texture, and stability of emulsions are strongly influenced by the character-istics of any flocs formed (e.g., their number, size, flexibility, and packing) (Walstra 1993a).The structure and properties of flocs are mainly determined by the nature of the colloidal andhydrodynamic interactions between the droplets, but they also depend on the mechanismresponsible for the droplet collisions (i.e., Brownian motion, gravity, or mechanical agitation)(Bremer 1992, Evans and Wennerstrom 1994). Valuable insights into the relationship be-tween these parameters and floc characteristics have been obtained from a combination ofcomputer simulations and experimental measurements.

7.4.3.1. Influence of Colloidal Interactions on Floc Structure

When the attraction between the droplets is relatively strong compared to the thermal energy,the flocs formed have extremely open structures in which each droplet is only linked to twoor three of its neighbors (Figure 7.13). This type of open structure is formed because thedroplets “stick” firmly together at the point where they first come into contact and are unableto undergo any subsequent structural rearrangements (Bremer 1992). As a consequence, adroplet which encounters a floc cannot move very far into its interior before becomingattached to another droplet. This type of floc is characterized by a tenuous structure whichtraps large amounts of continuous phase within it. The volume fraction of particles in sucha floc may be as low as 0.13 depending on its size and the strength of the interactions(Dickinson 1992). When the attraction between the droplets is relatively weak compared tothe thermal energy, the droplets do not always stick together after a collision and they maybe able to roll around each other after sticking together. Thus a droplet which encounters afloc is able to penetrate closer to its center and flocs are able to undergo structural rearrange-ments, which means that the droplets can pack more closely together (Bijsterbosch et al.1995). This type of floc is characterized by a more compact structure which traps less of thecontinuous phase. The volume fraction of the particles in this type of floc may be as high as0.63, which is close to the value for random packing of monodisperse particles (Dickinson1992).

FIGURE 7.13 Schematic diagram of some of the different types of structures that flocs may form.


FIGURE 7.14 Examples of two-dimensional structures that can be used to model flocs.

7.4.3.2. Use of Fractal Geometry to Describe Floc Structure

Fractal geometry has proven to be an extremely valuable tool for characterizing the structuralorganization of droplets in flocs (Bremer 1992, Peleg 1993, Walstra 1996a). Fractal geometryis applicable to systems which exhibit a phenomenon known as self-similarity (i.e., havestructures which appear similar when observed at different levels of magnification) (Figure7.14). There are no truly fractal objects in nature because all real objects have a definite upperand lower size. Nevertheless, many natural objects do show self-similarity over a number oflevels of magnification and can therefore be described by fractal geometry (Peleg 1993).Despite their extremely complex structures, these fractal objects can be described by a singleparameter (D) known as the fractal dimension.

The concept of a fractal dimension is best illustrated by considering the model two-dimensional fractal flocs shown in Figure 7.14. The floc with the “string-of-beads” typestructure has a fractal dimension of one (D = 1), because increasing the level of magnificationby a factor of X changes the number of particles per floc by a factor of X1 (i.e., N ∝ X1). Thefloc which contains the array of closely packed particles has a fractal dimension of two (D= 2), because increasing the level of magnification by a factor of X changes the number ofparticles per floc by a factor of X2 (i.e., N ∝ X2). These are the two extreme values of thefractal dimension of a two-dimensional structure. There are many other types of floc whichhave self-similar structures with intermediate fractal dimensions (Figure 7.14). The numberof particles in these flocs is described by the relationship N ∝ XD, where D is a nonintegervalue between 1 and 2. The closer the value is to one, the more tenuous is the floc structure,and the closer it is to two, the more compact. The fractals drawn in Figure 7.14 are charac-terized by a lower cutoff length (r) which is equal to the radius of the particles and an uppercutoff length (R) which is equal to the radius of the floc. The concept of a fractal dimensiontherefore only applies on length scales between these limits. The number of particles in afractal floc is given by the relationship N = (R/r)D.

In nature, flocs are three-dimensional structures, and so D ranges from 1 to 3: the higherthe fractal dimension, the more compact the floc structure (Peleg 1993). The volume fractionof droplets in a three-dimensional floc is given by (Bremer 1992):

φF

DRr

=

−3

(7.26)

This equation indicates that φF decreases as the size of a floc increases (because D < 3) and

that the concentration of droplets in a fractal floc decreases as one moves outward from its


interior. The fractal dimension of flocs can be determined using a variety of experimentaltechniques, including rheometry, light scattering, and microscopy (Bremer 1992). Thus it ispossible to quantify the influence of different parameters on the structure of the flocs formed(e.g., pH, ionic strength, and temperature).

It should be mentioned that fractal flocs are only kinetically stable structures (Evans andWennerstrom 1994). When there is a strong attraction between the droplets, the most ther-modynamically stable arrangement would be one in which the droplets formed a closelypacked structure, as this would maximize the number of favorable attractive interactions. Itis because there is a large activation energy associated with the rearrangement of dropletswithin a floc, which means that it retains its fractal structure.

7.4.3.3. Influence of Floc Structure on Emulsion Properties

The structure and properties of flocs have a pronounced effect on the stability and rheologicalproperties of emulsions. An emulsion that contains flocculated droplets has a higher viscositythan one that contains the same concentration of unflocculated droplets. This is because theeffective volume fraction of a floc is greater than the sum of the volume fractions of theindividual droplets due to the presence of the continuous phase trapped within it (Dickinsonand Stainsby 1982). The viscosity of emulsions usually increases as the floc structure be-comes more tenuous for the same reason. Emulsions that contain flocculated droplets exhibitpronounced shear thinning behavior (i.e., the viscosity decreases as the shear rate or sheartime increases) (Chapter 8). Shear thinning occurs for two reasons: (1) the flocs are deformedand become aligned with the shear field, which decreases their resistance to flow, and (2) theflocs are disrupted by the shear forces, which decreases their effective volume fraction(Figure 7.15) (Bujannunez and Dickinson 1994, Bower et al. 1997). The ease with which theflocs in an emulsion are deformed and disrupted decreases as the number and strength of theattractive interactions between the droplets increase (Uriev 1994, Pal 1996). Thus one wouldexpect that a higher shear stress would be required to disrupt flocs in emulsions that containdroplets that are strongly bound to each other than one in which they are only weakly bound.Once the shearing forces are removed from an emulsion, the bonds between the droplets mayreform. The rate at which this process occurs and the type of structures formed often havean important influence on the quality of food emulsions.

In some cases, shearing an emulsion can actually promote droplet flocculation because theefficiency and frequency of collisions between the droplets are increased (Spicer and Pratsinis1996). Thus an emulsion which is stable under quiescent conditions may become flocculatedwhen it is subjected to shear forces (Dalgleish 1996a). This type of emulsion initially showsshear thickening behavior because the formation of flocs leads to an increase in viscosity, but

FIGURE 7.15 Schematic diagram of the events that occur during the shearing of a flocculatedemulsion and their effect on the emulsion viscosity.


at higher shear rates these flocs may become deformed and disrupted, which leads to adecrease in viscosity (Pal 1996). Emulsions may therefore exhibit complex rheologicalbehavior depending on the nature of the colloidal interactions between the droplets.

At sufficiently high dispersed-phase volume fractions, flocculation leads to the formationof a three-dimensional network of aggregated droplets which extends throughout the emul-sion (Bijsterbosch et al. 1995). This type of system is referred to as a particle gel and can becharacterized by a yield stress which must be exceeded before the emulsion will flow (Pal1992, 1996; Pal et al. 1992; Tadros 1994, 1996). The value of the yield stress increases asthe dispersed-phase volume fraction increases and as the strength of the attractive forcesbetween the droplets increases (Pal 1996). The minimum dispersed-phase volume fractionrequired to form a particle gel decreases as the structure of the flocs becomes more tenuous(i.e., D → 1) because this type of structure is able to fill the available space more effectively(Figure 7.6). Thus two emulsions may have exactly the same dispersed-phase volume frac-tions, yet one could be a low-viscosity liquid and the other a viscoelastic gel, depending onthe nature of the colloidal interactions between the droplets.

Flocculation also affects the stability of emulsions to creaming. In dilute emulsions,flocculation increases the creaming rate because the effective size of the particles in theemulsion is increased, which more than compensates for the decrease in ∆ρ (Equation 7.9).In concentrated emulsions, the droplets are usually prevented from creaming because theformation of a three-dimensional network of aggregated droplets prevents them from moving(Figure 7.6).

7.4.4. Experimental Measurement of Flocculation

A wide variety of experimental techniques have been developed to monitor the extent and rateof flocculation in emulsions. The most direct method is to observe the emulsion through anoptical or electron microscope (Chapter 10). The association of droplets with one another canbe determined subjectively by eye or more quantitatively by transferring an image of theemulsion to a computer and using image analysis techniques (Jokela et al. 1990, Mikula 1992,Brooker 1995). The major drawback of most microscopic techniques is that the delicatestructure of the flocs may be disturbed by the procedures used to prepare the samples forobservation. In addition, in concentrated emulsions it is often difficult to tell whether twodroplets are flocculated or just in close proximity.

Flocculation can be monitored indirectly by measuring the change in the particle sizedistribution with time using a particle-sizing instrument (e.g., light scattering, ultrasonicspectrometry, or electrical pulse counting) (Chapter 10). These instruments often provide afairly qualitative indication of the extent of flocculation in an emulsion, because it isdifficult to specify the physical properties of a floc that are needed to mathematicallyconvert the experimental measurements into a true particle size distribution. For example,light-scattering techniques usually assume that the particles in an emulsion are isolatedhomogeneous spheres, whereas flocs are inhomogeneous and nonspherical aggregates thatcontain many individual droplets, and therefore the theory used to interpret the experimentaldata is no longer valid.

In many systems, it is important to determine whether the increase in droplet size is causedby flocculation, coalescence, or Ostwald ripening. The simplest method of achieving this isto alter the emulsion in a way which would be expected to break down any flocs that arepresent. If there are no flocs present, the particle size will not change after the alteration, butif there are flocs present, there will be a decrease in the particle size. A variety of methodsare available for breaking down flocs: (1) altering solvent conditions, such as pH, ionicstrength, dielectric constant, or temperature; (2) applying mechanical agitation, such asstirring or sonication; and (3) adding a more surface-active agent, such as a small-molecule


surfactant, which displaces the original emulsifier from the droplet interface and which doesnot cause flocculation itself. The choice of a suitable method depends on the nature of theemulsion and in particular on the type of emulsifier used to stabilize the system. In emulsionswhere the interfacial membrane consists of proteins that are held together by extensiveintermolecular disulfide bonds, it may be necessary to use a combination of mercaptoethanol(to break the disulfide bonds) and a small-molecule surfactant (to displace the proteins).

As with microscopy, the preparation of samples for analysis using particle-sizing instru-ments often disturbs the structures of the flocs. For example, emulsions often have to bediluted before they can be analyzed, which can cause disruption of the flocs, particularlywhen there is only a weak attraction between the droplets. In addition, it is important to carryout dilution using a solvent that has similar properties to the continuous phase in which thedroplets were originally dispersed (e.g., ionic strength, pH, and temperature); otherwise, thefloc structure may be altered. The dispersion of droplets by stirring may also cause some ofthe flocs to break down, particularly when they are only held together by weak attractiveforces. Many of these problems can be overcome using modern particle-sizing techniquesbased on ultrasonics, nuclear magnetic resonance, or dielectric spectrometry, because theycan be used to analyze concentrated emulsions without the need for any sample preparation(Chapter 10).

Flocculation causes an increase in the viscosity of an emulsion and may eventually leadto the formation of a particle gel (Section 7.4.3). As a consequence, flocculation can oftenbe monitored by measuring the change in viscosity or shear modulus of an emulsion (Groverand Bike 1995, Tadros 1996, Pal 1996). An indication of the strength of the attractive forcesbetween flocculated droplets can be obtained from measurements of the viscosity or shearmodulus versus shear stress (Pal 1996). As the shear stress is increased, the flocs becomedeformed and disrupted, so that the viscosity or shear modulus decreases. The stronger theattractive forces between the flocculated droplets, the higher the shear stress needed to disruptthem. Another indirect measurement of the degree of flocculation in an emulsion is todetermine the rate at which droplets cream or sediment (Luyten et al. 1993, Basaran et al.1998). As mentioned earlier, flocculation may either increase or decrease the creaming ratedepending on the droplet concentration and the structure of the flocs formed. Experimentalmethods that can be used to monitor creaming and sedimentation were discussed in Section7.3.3.

7.5. COALESCENCE

Coalescence is the process whereby two or more liquid droplets merge together to form asingle larger droplet (Figure 7.16). It is the principal mechanism by which an emulsion movestoward its most thermodynamically stable state because it involves a decrease in the contactarea between the oil and water phases (Section 7.2). Coalescence causes emulsion dropletsto cream or sediment more rapidly because of the increase in their size. In oil-in-water

FIGURE 7.16 Droplet coalescence eventually leads to the complete separation of the oil and aqueousphases.


emulsions, coalescence eventually leads to the formation of a layer of oil on top of thematerial, which is referred to as oiling off. In water-in-oil emulsions, it leads to the accumu-lation of water at the bottom of the material. For these reasons, an understanding of the factorswhich influence coalescence is important to food manufacturers attempting to create productswith extended shelf lives.

7.5.1. Physical Basis of Coalescence

Coalescence is the result of the liquid within two or more emulsion droplets coming intomolecular contact (Evans and Wennerstrom 1994, Kabalnov and Wennerstrom 1996, Walstra1996a,b). Thus this process can only occur when the droplets are close to each other and theinterfacial membranes are disrupted. The fact that the droplets must be in close contact meansthat coalescence is much more dependent on short-range forces, and therefore on the precisemolecular details of a system, than either gravitational separation or flocculation. The rate atwhich coalescence proceeds, and the physical mechanism by which it occurs, is thereforehighly dependent on the nature of the emulsifier used to stabilize the system. For this reason,our knowledge of coalescence is much less well developed than that of the other major formsof emulsion instability. Even so, a combination of theoretical and experimental studies hasled to a fairly good understanding of the major factors which influence coalescence in anumber of important systems. In general, the susceptibility of droplets to coalescence isdetermined by the nature of the forces that act on and between the droplets (i.e., gravitational,colloidal, hydrodynamic, and mechanical forces) and the resistance of the droplet membraneto rupture. It is useful to distinguish between two different types of situations which can leadto coalescence in food emulsions.

7.5.1.1. Coalescence Induced by Collisions

In this situation, coalescence occurs immediately after two or more droplets encounter eachother during a collision. It is the dominant form of coalescence in emulsions in which thedroplets are free to move around and collide with each other. The movement of the dropletsmay be induced by Brownian motion, gravity, or applied mechanical forces (Section 7.4.1.1).The rate at which this type of coalescence occurs is governed by many of the same factorsas flocculation (i.e., the collision frequency and efficiency) (Section 7.4.1). However, thecollision efficiency is now determined by the probability that the interfacial membranessurrounding the droplets will be ruptured during a collision. In the absence of an emulsifier,the coalescence collision efficiency is high (EC → 1), because there is little resistance to thedroplets merging together. In the presence of emulsifier, the collision efficiency may beextremely low (EC → 0), because of short-range repulsive interactions between the dropletsor because of the low probability that the interfacial membrane will be ruptured. For thisreason, this type of coalescence is only significant in systems in which there is insufficientemulsifier present to completely saturate the surface of the droplets or during homogeniza-tion, where emulsifiers may not have sufficient time to cover the droplet surfaces before adroplet–droplet collision occurs (Chapter 6). It may also be important in emulsions which aresubjected to high shear forces, because the impact forces that act on the droplets as the resultof a collision may then be sufficient to cause disruption of the interfacial membranes.

7.5.1.2. Coalescence Induced by Prolonged Contact

In this situation, coalescence occurs spontaneously after the droplets have been in contact fora prolonged period. This type of coalescence is therefore important in emulsions which havehigh droplet concentrations, which contain flocculated droplets, or which contain dropletsthat have accumulated at the top (or bottom) of the sample due to creaming (or sedimenta-


tion). It is no longer convenient to describe the rate at which coalescence occurs in terms ofa collision frequency and efficiency, because the droplets are in contact with each other.Instead, it is more useful to characterize this type of coalescence in terms of a coalescencetime, that is, the length of time the droplets remain in contact before coalescence occurs. Inpractice, all of the droplets in an emulsion do not usually coalesce after the same time, andtherefore it is more appropriate to define an average coalescence time or to stipulate a rangeof times over which the majority of the droplets coalesce.

One of the most important factors that determine the rate at which this process occurs isthe thickness of the layer of continuous phase which normally separates the droplets (Hunter1989). This thickness is governed mainly by the dependence of the interdroplet pair potentialon droplet separation (Chapter 3). Droplets tend to exist at a separation where the interdropletpair potential is at a minimum, although they may be forced closer together in the presenceof external forces, such as gravity or centrifugation. The coalescence stability of an emulsionis usually reduced as the thickness of the layer of continuous phase separating the dropletsis decreased (Section 7.5.2).

This type of coalescence is the most important in food emulsions (under quiescent condi-tions), because in most systems there is usually enough emulsifier present to make thedisruption of the interfacial membranes during a collision a highly improbable event.

7.5.1.3. Coalescence as the Result of “Hole” Formation

In many emulsions in which coalescence is observed, one would not expect it to occur froman examination of the interdroplet pair potential, because of the extremely high steric repul-sion that arises when two droplets closely approach each another (Section 3.5). The interdropletpair potential, which describes the dependence of the colloidal interactions between theemulsion droplets on separation, assumes that the system is at thermodynamic equilibrium.It therefore predicts that droplets will remain indefinitely in a potential energy minimum. Inpractice, the stability of an emulsion to coalescence is often governed by nonequilibriumprocesses associated with the dynamic events that occur on the molecular level (Evans andWennerstrom 1994, Kabalnov and Wennerstrom 1996).

It is widely accepted that coalescence occurs as the result of the formation of a hole whichextends across the interfacial membranes surrounding the droplets (Figure 7.17). Once thehole is formed, the liquid in the droplets rapidly flows through it and the droplets mergetogether to form a single larger droplet. The coalescence rate therefore depends on thelikelihood that these holes will form in the interfacial membranes.

Holes may be formed in an interfacial membrane in a variety of different ways, dependingon type of emulsifier used and the environmental conditions. They form spontaneously insurfactant monolayers due to thermal fluctuations of their shape (Section 7.5.1.4). They mayalso be formed as the result of the chemical breakdown of an emulsifier over time (e.g., aprotein or polysaccharide may be cleaved into smaller fragments due to enzymic or chemicalhydrolysis). These fragments might not be surface active or they may not form a goodprotective membrane, which leads to the formation of a hole. A hole may also be createdwhen emulsifiers are displaced from the surface of emulsion droplets by more surface-activecomponents which do not form membranes resistant to rupture (e.g., biopolymers may bedisplaced by high concentrations of alcohol). Which of these mechanisms is important in aparticular system depends on the composition and microstructure of the emulsion, as well asenvironmental conditions such as pH, ionic strength, ingredient interactions, temperature, andmechanical agitation (Section 7.5.3). Quantitative predictions of the rate at which coalescenceproceeds in real emulsions are extremely difficult because of the complex nature of theprocesses involved, the difficulty in developing theories to describe these processes, and thelack of experimental techniques to provide the information required to test these theories. The


(a)

(b)

FIGURE 7.17 Coalescence of droplets depends on (a) film thinning and (b) film rupture. (a) As twodroplets approach each other, the liquid between them gets increasingly thinner and may continue tothin or may reach some equilibrium value. (b) Two droplets that are in contact may spontaneouslymerge because of the thermal fluctuation of their membranes, leading to hole formation.

most significant progress in this area has been made in understanding the origins of holeformation and coalescence in emulsions stabilized by small-molecule surfactants (see nextsection).

7.5.1.4. Coalescence in Surfactant-Stabilized Systems

The susceptibility of emulsions stabilized by small-molecule surfactants to coalescence isgoverned mainly by the characteristics of the interfacial membrane (i.e., the optimum curva-ture, interfacial tension, and interfacial rheology) (Section 4.5). Two different physical mecha-nisms have been proposed for coalescence in surfactant-stabilized emulsions (Evans andWennerstrom 1994):

1. The surface of an emulsion droplet stabilized by a small-molecule surfactant ishighly dynamic and its shape is constantly changing because of its thermal energy.As a consequence of these thermal fluctuations, bare patches which are not pro-tected by surfactant may be temporarily formed. If two of these patches occursimultaneously at a point where the droplets are in contact, then a hole may beformed through which the dispersed phase can flow (Figure 7.17). This mechanismis likely to be important in systems which have very low interfacial tensions andin which the membrane has a low resistance to bending and dilation.

2. The size of the droplets in emulsions is usually orders of magnitude greater thanthe size of the surfactant molecules, and therefore the interface can be consideredto be approximately planar, that is, to have zero curvature (H = 0). As a conse-quence, the surfactant membrane may not be at its optimum curvature (H0). Theformation of a hole across the layer of continuous phase which separates thedroplets depends on the development of a highly curved edge (Figure 7.18). If thecurvature of the edge is close to the optimum curvature of the surfactant membrane(H0 ≈ Hedge), then the formation of a hole is energetically favorable. On the other


FIGURE 7.19 Gibbs–Marangoni effect.

FIGURE 7.18 Coalescence may occur when the optimum curvature of a surfactant is similar to thatof the edge of the hole.

hand, if the optimum curvature of the surfactant is opposite to that of the edge, thenthe formation of a hole is energetically unfavorable. The dependence of holeformation on the optimum curvature of a membrane means that coalescence isrelated to the molecular geometry of the surfactant molecules (Section 4.5). Therelationship between the molecular geometry and coalescence stability of oil-in-water and water-in-oil emulsions is highlighted in Figure 7.18. As with the othercoalescence mechanism described above, hole formation depends on thermal fluc-tuations of the interfacial membrane so that it is able to change its curvature. Thecoalescence rate will therefore increase as the interfacial tension and rigidity of themembrane decrease.

An additional factor which contributes to the stability of surfactant-stabilized emulsions inwhich collision-induced coalescence is important is the Gibbs–Marangoni effect (Walstra1993b, 1996a,b). As two droplets approach each other, the liquid in the continuous phase isforced out of the narrow gap which separates them. As the liquid is squeezed out, it dragssome of the surfactant molecules along the droplet surface, which leads to the formation ofa region where the surfactant concentration on the surfaces of the two emulsion droplets islowered (Figure 7.19). This causes a surface tension gradient at the interface, which isenergetically unfavorable. The surfactant molecules therefore have a tendency to flow towardthe region of low surfactant concentration and high interfacial tension, dragging some of theliquid in the surrounding continuous phase along with them. This motion of the continuous


phase is in the opposite direction of the outward flow that occurs when it is squeezed frombetween the droplets, and therefore it opposes the movement of the droplets toward eachother and therefore increases their stability to coalescence.

7.5.2. Methods of Controlling Coalescence

The rate at which coalescence occurs is strongly dependent on the structure and dynamics ofthe interfacial membranes which surround the droplets. As a consequence, the most appro-priate method of controlling coalescence is highly dependent on the type of emulsifier usedto stabilize the system, as well as the prevailing environmental conditions. Even so, it ispossible to give some general advice about the most effective methods of avoiding coales-cence. These methods can be conveniently divided into two categories: those which preventextended close contact and those which prevent membrane disruption.

7.5.2.1. Prevention of Extended Close Contact

The coalescence rate can be decreased by reducing the length of time that droplets are in closecontact. The droplet contact time can be reduced in a number of different ways, includingdecreasing their collision frequency, ensuring that they do not flocculate, and preventingthem from forming a concentrated layer at the top (or bottom) of an emulsion due to creaming(or sedimentation).

Even when the droplets are in contact with each other for extended periods (e.g., when theyare in a floc or in a creamed layer), they can still be prevented from coalescing by ensuringthat they do not get too close. The probability of coalescence occurring increases as thethickness of the layer of continuous phase separating them decreases, because the thermalfluctuations in the membrane shape may then become large enough to form a hole whichextends from one droplet to another. The thickness of this layer is determined principally bythe magnitude and range of the various attractive and repulsive forces that act between thedroplets (Chapter 3). The coalescence stability can therefore be enhanced by ensuring thereis a sufficiently large repulsive interaction to prevent the droplets from coming into closecontact. This can be achieved in a number of ways, including varying the emulsifier type, pH,ionic strength, or temperature.

7.5.2.2. Prevention of Membrane Disruption

The formation of a hole in an interfacial membrane depends on the fluctuations in its shapecaused by thermal energy or applied mechanical forces (Evans and Wennerstrom 1994). Themagnitude of these fluctuations is governed by the interfacial tension and rheology of themembrane (Israelachvili 1992, Evans and Wennerstrom 1994). The lower the interfacialtension, the greater the magnitude of the fluctuations and therefore the greater the likelihooda hole will form. The higher the viscoelasticity, the greater the short-range steric repulsionbetween the droplets and the greater the damping of the membrane fluctuations. Conse-quently, coalescence becomes less likely as the interfacial tension or the viscoelasticity of themembrane increases (Dickinson 1992). The thickness of the droplet membranes is also likelyto influence the coalescence stability of emulsions. Droplets with thicker membranes wouldbe expected to provide the greatest stability because they are less likely to be ruptured andthey provide a greater steric repulsion between the droplets. For small-molecule surfactants,it is important to select an emulsifier that has an optimum curvature which does not favorcoalescence (Section 7.5.1.4).

The likelihood of a hole forming in a given region within a membrane increases as its areaincreases. Consequently, the larger the contact area between two droplets, the greater thecoalescence rate. The coalescence rate therefore increases as the size of the droplets in an


emulsion increases, or when the droplets become flattened against one another. Dropletflattening occurs in highly concentrated emulsions, in creamed layers, or when emulsions aresubjected to centrifugal forces (Dickinson and Stainsby 1982, Walstra 1996a). Large dropletsare more prone to flattening than smaller droplets because the interfacial forces which tendto keep them in a spherical shape are lower (Chapter 6). Large droplets are also more proneto collision-induced coalescence because the impact forces generated during a collision aregreater and the magnitude of the attractive forces between the droplets is larger. On the otherhand, increasing the size of the droplets decreases the frequency of the encounters betweendroplets, which may be the dominant effect in emulsions where the rate-limiting step is thecollision frequency.

7.5.3. Influence of Emulsifier Type and Environmental Conditions

Protein emulsifiers have been found to be extremely effective at providing protection againstcoalescence (Dickinson 1992). The main reason for this is that proteins are capable ofproducing emulsions with small droplet sizes, they provide strong repulsive forces betweendroplets (due to a combination of electrostatic and steric interactions), the interfacial tensionis relatively high, and they form membranes which are highly viscoelastic. Partially hydro-lyzed proteins are less effective at preventing coalescence because they tend to form thinner,less viscoelastic interfacial layers that are easier to rupture.

Coalescence of emulsions stabilized by small-molecule surfactants is largely governed bytheir ability to keep droplets apart, rather than the resistance of the droplet membrane torupture. Nonionic surfactants, such as the Tweens, do this by having polymeric hydrophilichead groups that provide a large steric overlap and hydration repulsion (McClements andDungan 1997). Ionic surfactants, such as SDS and fatty acids, achieve this mainly throughelectrostatic repulsion (Section 3.4). Nevertheless, this electrostatic repulsion is only appre-ciable when the aqueous phase has a low ionic strength. At high ionic strengths, the electro-static repulsion is screened by the counterions, so that the droplets come close together, andare prone to coalescence, because the membrane is easily disrupted due to its low interfacialviscosity and interfacial tension.

Food manufacturers often need to create emulsions with extended shelf lives, and so it isimportant for them to understand the influence of various types of processing and storageconditions on droplet coalescence. Under quiescent conditions, emulsions stabilized by milkproteins are fairly stable to coalescence, provided the droplets are completely liquid (Das andKinsella 1990). However, droplet coalescence may occur when the droplets are subjected toshear forces or brought into close contact for extended periods (e.g., in a creamed layer orconcentrated emulsion) (Dickinson and Williams 1994). The presence of small amounts oflow-molecular-weight surfactants in the aqueous phase greatly enhances the tendency ofemulsion droplets to coalesce during shearing (Chen et al. 1993, Dickinson et al. 1993b, Lipset al. 1993). Emulsion instability probably occurs because small-molecule surfactants areadsorbed to the interface and increase the mobility of the proteins, therefore increasing theease with which the interfacial layer is ruptured (Dickinson et al. 1993b). The stability ofemulsions to shear forces therefore depends on the structure and properties of the adsorbedinterfacial layer. Further study is needed to increase our understanding of the relationshipbetween the physicochemical properties of emulsions during shear and the properties of theadsorbed protein.

When an oil-in-water emulsion is frozen, only part of the water is initially crystallized andthe oil droplets are forced into the remaining liquid region (Sherman 1968, Berger 1976,Dickinson and Stainsby 1982). The ionic strength of this region is increased significantlybecause of the concentration of salts and other components. The combination of forcing thedroplets into a more confined space and altering the solvent conditions is often sufficient to


disrupt the droplet membranes and promote coalescence once an emulsion is thawed (Berger1976). In addition, the oil droplets may crystallize during the freezing process, which can leadto emulsion instability through partial coalescence (Section 7.6). The development of freeze–thaw stable emulsions is a major challenge to food scientists.

Coalescence may also be promoted when an emulsion is dried into a powder, becausedrying may disrupt the integrity of the interfacial layer surrounding the droplets (Young etal. 1993a), which leads to coalescence once the emulsion is reconstituted. The coalescencestability can often be improved by adding relatively high concentrations of protein or carbo-hydrates to the system prior to drying (Young et al. 1993a,b). These molecules form a thickinterfacial membrane around the droplets which is less prone to disruption during the dehy-dration process.

The centrifugation of an emulsion may also lead to extensive coalescence because thedroplets are forced together into a compact droplet-rich layer with sufficient force to flattenthe droplets and disrupt the membranes (Dickinson and Stainsby 1982).

The coalescence of droplets in an emulsion may also be influenced by various chemicalor biochemical changes that occur over time. Lipid oxidation leads to the development ofsurface-active reaction products which may be capable of displacing emulsifier moleculesfrom the droplet surface and thereby promoting coalescence (Coupland and McClements1996). Extensive enzymatic hydrolysis of proteins or polysaccharides may cause an interfa-cial layer to be disrupted, again promoting droplet coalescence. An understanding of thevarious chemical and biochemical factors that determine coalescence is therefore essential increating emulsions with extended shelf lives or that can be broken down under specificconditions.

7.5.4. Experimental Measurement of Droplet Coalescence

Experimental characterization of coalescence can be carried out using a variety of experimen-tal techniques, many of which are similar to those used to monitor flocculation (Section7.4.3). The most direct approach is to observe droplet coalescence using an optical micro-scope (Mikula 1992). An emulsion is placed on a microscope slide and the change in thedroplet size distribution is measured as a function of time, by counting the individual dropletsmanually or by using a computer with image-processing software. It is possible to observeindividual coalescence events using a sufficiently rapid camera, but these events are often soimprobable in food emulsions that they are difficult to follow directly (Dickinson 1992,Walstra 1996a).

An alternative microscopic method involves the observation of the coalescence of singleemulsion droplets at a planar oil–water interface (Dickinson et al. 1988). An oil droplet isreleased from a capillary tube into an aqueous phase and moves to the oil–water interface dueto gravity (Figure 7.20). The time taken for the droplet to merge with the interface after it hasarrived there is measured. The results from this type of experiment demonstrate that there aretwo stages to droplet coalescence: (1) a lag phase corresponding to film thinning, where thedroplet remains at the interface but no coalescence occurs, and (2) a coalescence phase wherethe membrane spontaneously ruptures and the droplets merge with the bulk liquid. Dropletsexhibit a spectrum of coalescence times because membrane rupture is a chance process.Consequently, there is an approximately exponential decrease in the number of uncoalesceddroplets remaining at the interface with time after the lag phase. The major disadvantages ofthis technique are that only droplets above about 1 µm can be observed and coalescence oftenoccurs so slowly that it is impractical to monitor it continuously using a microscope. Todetect coalescence over a reasonably short period, it is necessary to have relatively lowconcentrations of emulsifier at the surfaces of the droplets, which is unrealistic because thedroplets in food emulsions are nearly always saturated with emulsifier.


Droplet coalescence can also be monitored by measuring the time dependence of thedroplet size distribution using an instrumental particle-sizing technique, such as light scatter-ing, electrical pulse counting, or ultrasonics (Chapter 10). These techniques are fully auto-mated and provide a measurement of the size of a large number of droplets in only a fewminutes. Nevertheless, it is important to establish whether the increase in droplet size is dueto coalescence, flocculation, or Ostwald ripening. Coalescence can be distinguished fromflocculation by measuring the droplet size distribution in an emulsion, then changing theenvironmental conditions so that any flocs are broken down, and remeasuring the droplet sizedistribution (Section 7.4.3). If no flocs are present, the average droplet size remains constant,but if there are flocs present, it decreases. Coalescence is more difficult to distinguish fromOstwald ripening because they both involve an increase in the average size of the individualdroplets with time.

As mentioned earlier, studies of coalescence can take a considerable time to completebecause of the very slow rate of the coalescence process. Coalescence studies can be accel-erated by applying a centrifugal force to an emulsion so that the droplets are forced together:the more resistant the membrane is to disruption, the higher the centrifugation force it cantolerate or the longer the time it will last at a particular speed before membrane disruptionis observed (Sherman 1995). Alternatively, coalescence can be accelerated by subjecting theemulsions to high shear forces and measuring the shear rate at which coalescence is firstobserved or the length of time that the emulsion must be sheared at a constant shear ratebefore coalescence is observed (Dickinson and Williams 1994). Nevertheless, these acceler-ated coalescence tests may not always give a good indication of the long-term stability of anemulsion. For example, chemical or biochemical changes may occur in an emulsion whichis stored for a long period that eventually lead to coalescence, but they may not be detectedin an accelerated coalescence test. Alternatively, there may a critical force that is required tocause membrane rupture which is exceeded in a centrifuge or shearing device, but whichwould never be exceeded under normal storage conditions. As a consequence, these acceler-ated coalescence tests must be used with caution.

7.6. PARTIAL COALESCENCE

Partial coalescence occurs when two or more partly crystalline oil droplets come into contactand form an irregularly shaped aggregate (Figure 7.21). The aggregate partly retains the shapeof the droplets from which it was formed because the fat crystal network within the dropletsprevents them from completely merging together (Mulder and Walstra 1974, Boode 1992,Dickinson and McClements 1995, Walstra 1996a). Partial coalescence is particularly impor-tant in dairy products, because milk fat globules are partly crystalline over a fairly wide rangeof temperatures (Mulder and Walstra 1974, Walstra and van Beresteyn 1975). The applica-

FIGURE 7.20 Microscopic technique for monitoring coalescence.


tion of shear forces or temperature cycling to cream containing partly crystalline milk fatglobules can cause partial coalescence, which leads to a marked increase in viscosity (“thick-ening”) and subsequent phase separation (Van Boekel and Walstra 1981, Boode et al. 1991,Boode 1992). Partial coalescence is an essential process in the production of ice cream,whipped toppings, butter, and margarine (Dickinson and Stainsby 1982; Goff et al. 1987;Barford and Krog 1987; Barford et al. 1987, 1991; Moran 1994). Oil-in-water emulsions arecooled to a temperature where the droplets are partly crystalline and a shear force is appliedwhich leads to droplet aggregation via partial coalescence (Mulder and Walstra 1974). Inbutter and margarine, aggregation results in phase inversion (Moran 1994), whereas in icecream and whipped cream, the aggregated fat droplets form a network that surrounds the aircells and provides the necessary mechanical strength required to produce good stability andtexture (Barford et al. 1987, Goff 1993).

7.6.1. Physical Basis of Partial Coalescence

Partial coalescence is initiated when a solid fat crystal from one droplet penetrates into theliquid oil portion of another droplet (Boode 1992; Boode and Walstra 1993a,b; Boode et al.1993; Walstra 1996a). Normally, the crystal would be surrounded by the aqueous phase, butwhen it penetrates into another droplet, it is surrounded by liquid oil. This causes the dropletsto remain aggregated because it is energetically more favorable for a fat crystal to besurrounded by oil molecules than by water molecules (i.e., the fat crystal is wetted better byliquid oil than by water). Over time, the droplets merge more closely together because thisreduces the surface area of oil exposed to water (Figure 7.21).

Partial coalescence may occur immediately after two droplets come into contact with eachother, or it may occur after the droplets have been in contact for an extended period (Boode1992). It is affected by many of the same factors which influence normal coalescence,including contact time, collision frequency, droplet separation, colloidal and hydrodynamicinteractions, interfacial tension, and membrane viscoelasticity (Section 7.5.1). Nevertheless,there are also a number of additional factors which are unique to partial coalescence, the mostimportant being the fact that the oil phase is crystalline (Boode and Walstra 1993a,b; Walstra1996a).

The solid fat content (φSFC) is the percentage of fat that is crystalline at a particulartemperature, varying from 0% for a completely liquid oil to 100% for a completely solid fat.Partial coalescence only occurs in emulsions that contain partly crystalline droplets, becausea solid fat crystal from one droplet must penetrate into the liquid oil region of another droplet(Boode and Walstra 1993a,b). If the droplets were completely liquid, they would undergonormal coalescence, and if they were completely solid, they would undergo flocculationrather than partial coalescence because the droplets are not able to merge together. Increasingthe φSFC from 0% causes an initial increase in the partial coalescence rate until a maximum

FIGURE 7.21 Partial coalescence occurs between two droplets that are partly crystalline when acrystal from one droplet penetrates into the liquid portion of another droplet. With time, the dropletsmerge together.


FIGURE 7.23 Typical distributions of fat crystals found in milk fat globules.

FIGURE 7.22 Schematic diagram of the influence of SFC on the rate of partial coalescence inemulsions.

value is reached, after which the partial coalescence rate decreases (Boode and Walstra1993a,b). Consequently, there is a certain φSFC at which the maximum rate of partial coales-cence occurs (Figure 7.22). The solid fat content at which this maximum rate occurs dependson the morphology and location of the crystals within the droplets, as well as the magnitudeof the applied shear field (Boode 1992, Boode and Walstra 1993a,b).

The dimensions and location of the fat crystals within an emulsion droplet play an impor-tant role in determining its susceptibility to partial coalescence (Campbell 1991, Darling1982). The further a fat crystal protrudes into the aqueous phase, the more likely it is topenetrate another droplet and therefore cause partial coalescence (Darling 1982, Walstra1996a). The different types of partially crystalline fat globule commonly observed in milk arerepresented in Figure 7.23 (Walstra 1967). Milk fat usually crystallizes as small platelets thatappear needle-shaped when observed by polarized light microscopy (Walstra 1967, Boode1992). These crystals may be evenly distributed within the interior of the droplet (type N),located exclusively at the oil–water interface (type L), or a combination of the two (type M).The type of crystals formed depends on whether the nucleation is homogeneous, surfaceheterogenous, or volume heterogeneous, as well as the cooling conditions (Dickinson andMcClements 1995).

It is possible to alter the distribution of fat crystals within an emulsion droplet after theyhave been formed by carefully cycling the temperature (Boode et al. 1991, Boode 1992).When milk is cooled rapidly, there is a tendency for the fat crystals within the droplets to befairly evenly distributed (type N). Thermodynamically, it is actually more favorable for thesecrystals to be located at the oil–water interface rather than in the interior of the droplets;however, their movement to the interface is restricted because they are trapped within the fatcrystal network. If the emulsion is heated to a temperature where most (but not all) of thecrystals melt, the crystal network is broken down, and the remaining crystals move to the


FIGURE 7.24 Ostwald ripening involves the growth of large droplets at the expense of smaller ones.

interface unhindered by the crystal network. When the emulsion is cooled, these crystals actas nucleation sites and crystal growth is restricted to the droplet surface. Thus L-type dropletsare formed with large crystals located at the oil–water interface. This phenomenon is believedto be responsible for the increase in the rate of partial coalescence which occurs duringtemperature cycling of oil-in-water emulsions (Boode et al. 1991). The crystals at the inter-face in L-type droplets are larger and protrude farther into the aqueous phase and are thereforemore likely to cause partial coalescence (Boode 1992).

In other types of food emulsions, there may be different crystal structures within thedroplets than those shown in Figure 7.24. The crystal structure formed will depend on factorssuch as the chemical structure of the fat, the cooling rate, temperature cycling, the applicationof shear forces, droplet size distribution, the type of emulsifier used to stabilize the emulsiondroplets, and the presence of any impurities that can poison or catalyze crystal growth.Further work is needed to elucidate the relationship between the morphology and location ofcrystals within emulsion droplets and their propensity to undergo partial coalescence.

7.6.2. Methods of Controlling Partial Coalescence

7.6.2.1. Prevention of Extended Close Contact

Partial coalescence is more likely to occur in emulsions that contain droplets which remainin contact for extended periods (i.e., flocculated emulsions, concentrated emulsions, andcreamed layers). For example, experiments have shown that partial coalescence occurs morerapidly when oil droplets are located in a creamed layer than when they are freely suspendedin the aqueous phase (Boode 1992). This is because the contact time is greater and theinterdroplet separation is reduced. In emulsions that contain freely suspended droplets, therate of partial coalescence is proportional to the frequency of encounters between emulsiondroplets. Thus anything which increases the collision frequency will increase the rate ofpartial coalescence (provided it does not also reduce the collision efficiency). The precisenature of these effects depends on whether the droplet collisions are induced by Brownianmotion, shear, or gravity (Van Boekel and Walstra 1981, Boode et al. 1991). For example,many emulsions are stable to partial coalescence at rest, but are prone to “thickening” whensheared.

Partial coalescence can only occur when droplets get so close together that a fat crystalfrom one droplet protrudes into another droplet. Thus any droplet–droplet interaction whichprevents droplets from coming into close contact should decrease the rate of partial coales-cence (e.g., electrostatic, polymeric steric overlap, hydration, and hydrodynamic repulsion).On the other hand, any droplet–droplet interaction which causes the droplets to come intoclose contact should increase the rate of partial coalescence (e.g., van der Waals, hydropho-bic, depletion, mechanical, centrifugational, and gravitational forces).


The size of the droplets in an emulsion affects partial coalescence in a number of ways.The efficiency of partial coalescence can increase with droplet size because there is a greaterprobability of there being a crystal present in the contact zone between the droplets (Boode1992). On the other hand, increasing the droplet size decreases their collision frequency,which can lead to a decrease in partial coalescence with increasing size (McClements et al.1994). The influence of droplet size is therefore quite complex and depends on the rate-limiting step in the partial coalescence process.

7.6.2.2. Prevention of Membrane Disruption

Emulsifiers that form thicker and more viscoelastic films at the oil–water interface are moreresistant to penetration by fat crystals and so provide greater stability to partial coalescence(Van Boekel and Walstra 1981, Boode et al. 1991, Boode 1992). Consequently, the rate ofpartial coalescence is less for droplets stabilized by proteins than for those stabilized bysmall-molecule surfactants (Dickinson and McClements 1995). The chemical structure of theemulsifier molecules at the interface of the emulsion droplets has an important impact on thestability of a number of foods. The presence of phospholipids in dairy emulsions has beenobserved to decrease their stability to thickening under the influence of shear forces, whichhas been attributed to the displacement of protein molecules from the oil–water interface bythe more surface-active phospholipids, leading to the formation of an emulsifier film whichis more susceptible to penetration by fat crystals (Boode 1992). When an ice cream mix ordairy whipped topping is aged in the presence of small-molecule surfactants prior to coolingand shearing, the resulting product has improved texture and better stability (Goff et al. 1987;Barford and Krog 1987; Barford et al. 1987, 1991). This is because emulsifiers displace themilk proteins from the oil–water interface and so enhance the tendency for partial coalescenceto occur during subsequent cooling and shearing. For these reasons, small-molecule surfac-tants are often added to ice creams and dairy toppings to improve their physical character-istics. Van Boekel and Walstra (1981) calculated that only about one in a million encountersbetween droplets leads to partial coalescence. This suggests that the rate-limiting step ofpartial coalescence is the penetration of the emulsifier film by a crystal and subsequentnucleation, rather than the collision frequency.

7.6.2.3. Control of Crystal Concentration, Structure, and Location

The degree of partial coalescence in an emulsion can be regulated by controlling the concen-tration, structure, and location of the fat crystals within the droplets. The most effectivemethod of preventing partial coalescence is to ensure that all the droplets are either com-pletely liquid or completely solid. This can be achieved by carefully controlling the tempera-ture of the emulsion so that it is above or below some critical level. If it is not possible toalter the temperature, then the solid fat content could be controlled by selecting a fat with adifferent melting profile (Moran 1994). The susceptibility of an emulsion to partial coales-cence also depends on the morphology and the precise location of the crystals within theemulsion droplets. Thus two emulsions with the same solid fat content may have verydifferent stabilities because of differences in crystal structure. The structure of the crystalscan be controlled by varying the cooling rate at which they are formed or by selecting anappropriate fat source.

7.6.3. Experimental Measurement of Partial Coalescence

A variety of experimental techniques have been used to monitor the susceptibility of foodemulsions to partial coalescence (Dickinson and McClements 1995). The physical state offats is monitored using experimental techniques that utilize differences in the physicochemi-


cal properties of the solid and liquid phases (e.g., density, compressibility, birefringence,molecular mobility, or packing) or changes associated with the solid–liquid phase transition(e.g., absorption or release of heat). The physicochemical properties which are of mostinterest to food scientists are (1) the final melting point of the fat; (2) the variation of the solidfat content with temperature; (3) the morphology, interactions, and location of the crystalswithin the droplets; (4) the packing of the fat molecules within the crystals; and (5) theinfluence of droplet crystallization on the overall stability and rheology of the emulsion.

The variation of the solid fat content with temperature can be measured using a variety oftechniques, including density measurements, differential scanning calorimetry, nuclear mag-netic resonance, ultrasonic velocity measurements, and electron spin resonance (Dickinsonand McClements 1995). The technique used in a particular experiment depends on theequipment available, the information required, and the nature of the sample being tested. Theposition of fat crystals relative to an oil–water interface depends on the relative magnitudeof the oil–water, oil–crystal, and crystal–water interfacial tensions and is characterized by thecontact angle (Darling 1982, Campbell 1991). Darling (1982) has described a techniquewhich can be used to measure the contact angle between liquid oil, solid fat, and aqueousphases.

Two fats can have exactly the same solid fat content but very different physical character-istics because of differences in the crystal habit and spatial distribution of the crystals. Thelocation of crystals within a system and their crystal habit can be studied by polarized lightmicroscopy (Boode 1992, Kellens et al. 1992) or electron microscopy (Soderberg et al. 1989),depending on the size of the crystals. The packing of the molecules in the crystals can bedetermined by techniques which utilize the scattering or adsorption of radiation. X-raydiffraction and small-angle neutron scattering have been used to determine the long and shortspacings of the molecules in fat crystals (Hernqvist 1990, Cebula et al. 1992). Infrared andRaman spectroscopy have been used to obtain information about molecular packing via itseffect on the vibration of certain chemical groups in fat molecules (Chapman 1965). Eachpolymorphic form has a unique spectra which can be used to identify it. The polymorphicform of fat crystals can also be identified by measuring the temperature at which phasetransitions occur and the amount of heat absorbed/released using differential scanning calo-rimetry (McClements et al. 1993a,b).

Partial coalescence leads to an increase in the size of the particles in an emulsion, whichcan be followed by microscopic, light-scattering, electrical pulse counting, or ultrasonictechniques (Chapter 10). Ultimately, a food scientist is interested in the influence of dropletcrystallization on the bulk physicochemical properties of a food emulsion, such as its appear-ance, stability, and texture. Partial coalescence causes an increase in emulsion viscosity andmay eventually lead to the formation of a three-dimensional network of aggregated droplets,so that it can be monitored by measuring the increase in viscosity or shear modulus, as afunction of either time or temperature (Boode 1992, Boode and Walstra 1993a,b). Thestability of emulsion droplets to creaming or sedimentation is also influenced by partialcoalescence, which can be followed by the techniques described in Section 7.3.3.

7.7. OSTWALD RIPENING

Ostwald ripening is the process whereby large droplets grow at the expense of smaller onesbecause of mass transport of dispersed phase from one droplet to another through theintervening continuous phase (Figure 7.24) (Kabalnov and Shchukin 1992, Taylor 1995). Itis negligible in most food emulsions because the mutual solubilities of triacylglycerols andwater are so low that the mass transport rate is insignificant (Dickinson and Stainsby 1982).Nevertheless, it may be important in oil-in-water emulsions which contain more water-


soluble lipids (e.g., flavor oils or when the aqueous phase contains alcohol (e.g., creamliqueurs). In this type of system, the food manufacturer may have to consider methods forretarding the rate of Ostwald ripening.

7.7.1. Physical Basis of Ostwald Ripening

Ostwald ripening occurs because the solubility of the material in a spherical droplet increasesas the size of the droplet decreases (Dickinson 1992):

S r SV

RTrm( ) ( ) exp= ∞

2γ(7.27)

Here, Vm is the molar volume of the solute, γ is the interfacial tension, S(∞) is the solubilityof the solute in the continuous phase for a droplet with infinite curvature (a planar interface),and S(r) is the solubility of the solute when contained in a spherical droplet of radius r. Theincrease in solubility with decreasing droplet size means that there is a higher concentrationof solubilized material around a small droplet than around a larger one. The solubilizedmolecules therefore move from the smaller droplets to the larger droplets because of thisconcentration gradient. Once steady state has been achieved, the rate of Ostwald ripening isgiven by (Kabalnov and Shchukin 1992):

d r

dt

V S D

RTm

3 8

9=

∞γ ( )(7.28)

where D is the translation diffusion coefficient of the solute through the continuous phase.This equation indicates that the change in droplet size with time becomes more rapid as theequilibrium solubility of the molecules in the continuous phase increases.

Equation 7.28 assumes that the emulsion is dilute, whereas most food emulsions areconcentrated, and so the growth or shrinkage of a droplet cannot be considered to beindependent of its neighbors. Thus the growth rate must be modified by a correction factorthat takes into account the spatial distribution of the droplets (Kabalnov and Shchukin 1992).Equation 7.28 also assumes that the rate-limiting step is the diffusion of the solute moleculesthrough the continuous phase. Most food emulsions contain droplets that are surrounded byinterfacial membranes which may retard the diffusion of solute molecules, and in this casethe above equation must be modified (Kabalnov and Shchukin 1992):

d r

dt

S S

R Rm c

m c

33

4=

−+

π (7.29)

where Rm and Rc are the diffusion resistances of the membrane and the continuous phase,respectively:

RrD

RC

r D Cmm

cm

c c

= = ∞

∞

14 4 2π

δπ

,

,(7.30)

Here, δ is the thickness of the droplet membrane, Ci,∞ is the solubility of the solute in thespecified phase, and the subscripts m and c refer to the properties of the membrane and the


continuous phase, respectively. When the diffusion of the solute molecules through theinterfacial membrane is limiting, the growth rate of the droplet size is proportional to r 2 ratherthan r 3 (Kabalnov and Shchukin 1992).

7.7.2. Methods of Controlling Ostwald Ripening

There are a number of ways in which the rate of Ostwald ripening in emulsions can becontrolled.

7.7.2.1. Droplet Size Distribution

The rate of Ostwald ripening increases as the average size of the droplets in an emulsiondecreases because the solubility of the dispersed phase increases with decreasing radius. Theinitial rate also increases as the width of the particle size distribution increases (Kabalnov andShchukin 1992). Ostwald ripening can therefore be retarded by ensuring that an emulsion hasa narrow droplet size distribution and that the droplets are fairly big. Nevertheless, there maybe other problems associated with having relatively large droplets in an emulsion, such asaccelerated creaming, flocculation, or coalescence.

7.7.2.2. Solubility

The greater the equilibrium solubility of the dispersed phase in the continuous phase, thefaster the rate of Ostwald ripening (Equation 7.28). Ostwald ripening is therefore extremelyslow in oil-in-water emulsions that contain lipids which are sparingly soluble in water, suchas long-chain triacylglycerols, but increases as the lipid molecules become smaller and morepolar (Dickinson and Stainsby 1988). Certain substances are capable of increasing the watersolubility of lipids in water and are therefore able to enhance the rate of Ostwald ripening(e.g., alcohols or surfactant micelles) (McClements et al. 1993b, Weiss et al. 1996, Couplandet al. 1997). Ostwald ripening could therefore be retarded by excluding these substances fromthe emulsion or by using lipids with a low water solubility.

7.7.2.3. Interfacial Membrane

The rate of Ostwald ripening increases as the interfacial tension increases (Equation 7.28).Consequently, it is possible to retard its progress by using an emulsifier which is highlyeffective at reducing the interfacial tension (Kabalnov et al. 1995). The mass transport ofmolecules from one droplet to another depends on the rate at which the molecules diffuseacross the interfacial membrane (Kabalnov and Shchukin 1992). It may therefore be possibleto retard Ostwald ripening by decreasing the diffusion coefficient of the dispersed phase inthe membrane or by increasing the thickness of the membrane. Little work has been carriedout in this area; however, it may prove to be an extremely useful means of controlling thestability of some food emulsions.

7.7.2.4. Droplet Composition

The rate of Ostwald ripening is particularly sensitive to the composition of emulsion dropletswhich contain chemical substances with different solubilities in the continuous phase (Kabalnovand Shchukin 1992, Dickinson and McClements 1995, Arlauskas and Weers 1996). Consideran oil-in-water emulsion that contains droplets comprised of two different types of organicmolecules: Mlow has a low water solubility and Mhigh has a high water solubility. The diffusionof Mhigh molecules from the small to the large droplets occurs more rapidly than the Mlow

molecules. Consequently, there is a greater percentage of Mhigh in the larger droplets than inthe smaller droplets. Differences in the composition of emulsion droplets are thermodynami-


cally unfavorable because of the entropy of mixing: it is entropically more favorable to havethe two oils distributed evenly throughout all of the droplets rather than concentrated inparticular droplets. Consequently, there is a thermodynamic driving force that operates inopposition to the Ostwald ripening effect. After a certain time, the driving force for Ostwaldripening (differences in droplet size) is exactly compensated for by the driving force for theentropy of mixing (differences in droplet composition), and so the size and composition ofthe droplets remain constant (Kabalnov and Shchukin 1992). It is therefore possible to controlthe rate of Ostwald ripening in oil-in-water emulsions by using an oil phase which containsa mixture of lipids with different water solubilities.

7.7.3. Experimental Measurement of Ostwald Ripening

Methods of monitoring Ostwald ripening are fairly similar to those used to monitor dropletcoalescence (i.e., techniques that measure changes in droplet size distribution with time)(Section 7.5.4). If the droplets are sufficiently large (>1 µm), then optical microscopy can beused (Kabalnov and Shchukin 1992); otherwise, instrumental particle-sizing techniques, suchas light scattering, electrical conductivity, or ultrasonics, can be used. Nevertheless, it is oftendifficult to directly distinguish between coalescence and Ostwald ripening using these par-ticle-sizing techniques because both instability mechanisms lead to an increase in the averagesize of the droplets over time.

7.8. PHASE INVERSION

Phase inversion is the process whereby a system changes from an oil-in-water emulsion toa water-in-oil emulsion or vice versa (Figure 7.25). Phase inversion is an essential step inthe manufacture of a number of important food products, including butter and margarine(Mulder and Walstra 1974, Dickinson and Stainsby 1982, Moran 1994). In other foods,phase inversion is undesirable because it has an adverse effect on their appearance, texture,stability, and taste. In these products, a food manufacturer wants to avoid the occurrence ofphase inversion.

7.8.1. Physical Basis of Phase Inversion

Phase inversion is usually triggered by some alteration in the composition or environmentalconditions of an emulsion (e.g., dispersed-phase volume fraction, emulsifier type, emulsifierconcentration, solvent conditions, temperature, or mechanical agitation) (Shinoda and Friberg1986, Dickinson 1992, Campbell et al. 1996). Only certain types of emulsion are capable of

FIGURE 7.25 Phase inversion involves the conversion of an oil-in-water emulsion to a water-in-oilemulsion or vice versa.


undergoing phase inversion, rather than being completely broken down into their componentphases. These emulsions are capable of existing in a kinetically stable state after the phaseinversion has taken place. It is usually necessary to agitate an emulsion during the phaseinversion process; otherwise it will separate into its component phases.

The physicochemical basis of phase inversion is believed to be extremely complex, involv-ing aspects of flocculation, coalescence, partial coalescence, and emulsion formation. At thepoint where phase inversion occurs, which is often referred to as the “balance point,” thesystem may contain regions of oil-in-water emulsion, water-in-oil emulsion, multiple emul-sion, and bicontinuous phases. Phase inversion in food emulsions can be conveniently di-vided into two different categories according to its origin.

7.8.1.1. Surfactant-Induced Phase Inversion

This type of phase inversion occurs in emulsions which are stabilized by small-moleculesurfactants and is caused by changes in the molecular geometry of the surfactant molecules(Shinoda and Friberg 1986, Salager 1988, Evans and Wennerstrom 1994). An emulsionstabilized by a nonionic surfactant undergoes a transition from an oil-in-water emulsion to abicontinuous system to a water-in-oil emulsion on heating because of the progressive dehy-dration of the head groups (Lehnert et al. 1994). This process is characterized by a phaseinversion temperature, which is governed by the molecular geometry of the surfactant mol-ecules (Davis 1994b). An oil-in-water emulsion stabilized by an ionic surfactant exhibits asimilar kind of behavior when the concentration of electrolyte in the aqueous phase isincreased (Salager 1988, Binks 1993). Increasing the ionic strength causes the system toundergo a transition from an oil-in-water emulsion to a bicontinuous system to a water-in-oil emulsion, because the electrical charge on the surfactant head groups is progressivelyscreened by the counterions (Section 4.5). Surfactant-induced phase inversions are reversiblebecause when the emulsion is cooled or the ionic strength is decreased, the water-in-oilemulsion will revert back to an oil-in-water emulsion. Nevertheless, it is usually necessaryto continuously agitate the system during this process, or else it will separate into theindividual oil and water phases. Even though the phase inversion is reversible, there is oftenan effect which is similar to that of supercooling (i.e., the temperature at which the phaseinversion occurs on heating is different from that on cooling) (Dickinson 1992, Vaessen andStein 1995). This is because there is an activation energy which must be overcome before asystem can be transformed from one state to another. The principal factor determiningreversible phase inversion is the molecular geometry of the surfactant molecules and itsdependence on environmental conditions.

7.8.1.2. Fat Crystallization–Induced Phase Inversion

When an oil-in-water emulsion that contains completely liquid droplets is cooled to a tem-perature where the droplets are partly crystalline and then sheared, it undergoes a phaseinversion to a water-in-oil emulsion (Mulder and Walstra 1974, Dickinson and Stainsby1982). The principal cause of this type of phase inversion is partial coalescence of thedroplets, which leads to the formation of a continuous fat crystal network which traps waterdroplets within it. This is one of the principal manufacturing steps in the production ofmargarine and butter (Dickinson and Stainsby 1982). When the emulsion is heated to atemperature where the fat crystals melt, the emulsion breaks down because the water dropletsare released and sediment to the bottom of the sample, where they coalesce with other waterdroplets. This is clearly seen when one melts margarine or butter and then cools it back tothe original temperature: the system before and after heating is very different. This type ofphase inversion depends mainly on the crystallization of the fat and on the resistance of thedroplets to partial coalescence (see Section 7.6.2).


7.8.2. Methods of Controlling Phase Inversion

The propensity for phase inversion to occur in an emulsion can be controlled in a number ofways.

7.8.2.1. Dispersed Volume Fraction

When the dispersed-phase volume fraction of an emulsion is increased while all the otherexperimental variables are kept constant (e.g., emulsifier type, emulsifier concentration,temperature, shearing rate), a critical volume fraction (φc) is reached where the system eitherundergoes a phase inversion or completely breaks down so that the excess dispersed phaseforms a layer on top of the emulsion. There is usually a range of volume fractions over whichan emulsion can exist as either a water-in-oil emulsion or an oil-in-water emulsion (Dickinson1992). Within this region, the emulsion can be converted from one state to another by alteringsome external property, such as the temperature or shear rate. From geometrical packingconsiderations, it has been estimated that this range extends from 1 – φcp < φ < φcp, where φcp

refers to the volume fraction when the droplets are packed closely together without beingdistorted. In practice, factors other than simple geometric considerations will influence thisrange, including the fact that the droplets can become deformed and the chemical structureof the emulsifier used. Phase inversion can therefore be prevented by ensuring that the dropletconcentration is kept below φcp.

7.8.2.2. Emulsifier Type and Concentration

The most important factor determining the susceptibility of an emulsion to surfactant-inducedphase inversion is the molecular geometry of the surfactant used to stabilize the droplets(Evans and Wennerstrom 1994). Surfactant-stabilized emulsions undergo a phase inversionwhen some change in the environmental conditions causes the optimum curvature of thesurfactant monolayer to tend toward zero (e.g., temperature, ionic strength, or the presenceof a co-surfactant) (Section 4.5). These types of surfactants usually have an intermediate HLBnumber or are electrically charged. Alternatively, mixtures of two different types of surfac-tants can be used, one to stabilize the oil-in-water emulsion and the other the water-in-oilemulsion. The point at which phase inversion occurs is then sensitive to the overall concen-tration and ratio of the surfactants used (Dickinson 1992). Emulsions stabilized by proteinsdo not exhibit this type of phase inversion because proteins are incapable of stabilizing water-in-oil emulsions.

The emulsifier type and concentration are also extremely important in determining thestability of emulsions to fat crystallization–induced phase inversion. Emulsifiers that formthick viscoelastic membranes are more likely to protect an emulsion from this type of phaseinversion because they retard partial coalescence (Section 7.6). It is therefore extremelyimportant to select an emulsifier which exhibits the appropriate behavior over the experimen-tal conditions that a food emulsion experiences during its lifetime.

7.8.2.3. Mechanical Agitation

It is often necessary to subject an emulsion to a high shearing force to induce phase inversionin the region where it can possibly exist as either an oil-in-water or water-in-oil emulsion. Thehigher the shearing force, the more likely phase inversion is to occur.

7.8.2.4. Temperature

Increasing or decreasing the temperature of emulsions is one of the most important means ofinducing phase inversion. The mechanism by which this process occurs depends on whether


the phase inversion is induced by surfactant changes or crystallization. Cooling an oil-in-water emulsion to a temperature where the oil partly crystallizes and shearing causes fatcrystallization–induced phase inversion. On the other hand, heating an oil-in-water emulsionstabilized by a surfactant may cause surfactant-stabilized phase inversion above the phaseinversion temperature.

Fat crystallization–induced phase inversion is the most important type in the food industrybecause it is an essential step in the manufacture of butter and margarine (Dickinson andStainsby 1982). Surfactant-induced phase inversion may be important in emulsions stabilizedby nonionic surfactants that must be heated to high temperatures. In these systems, it isimportant for the food manufacturer to ensure that the phase inversion temperature of theemulsifier is above the highest temperature that the emulsion will experience during process-ing, storage, and handling.

7.8.3. Experimental Measurement of Phase Inversion

Phase inversion can be monitored using a variety of experimental techniques (Lehnert et al.1994). When an emulsion changes from the oil-in-water to the water-in-oil emulsion type, orvice versa, there is usually a significant change in emulsion viscosity. This is because theviscosity of an emulsion is governed principally by the viscosity of the continuous phase(which is different for oil and water) and the dispersed-phase volume fraction (which mayalso be altered) (Chapter 8). An oil-in-water emulsion has an aqueous continuous phase andso its electrical conductivity is much greater than that of a water-in-oil emulsion (Lehnert etal. 1994, Keikens 1997). Thus there is a dramatic reduction in the electrical conductivity ofan oil-in-water emulsion when phase inversion occurs. Information about the process ofphase inversion may also be obtained by monitoring the emulsion under a microscope ormeasuring the size of the droplets using a particle-sizing technique. Measurements of thetemperature dependence of the interfacial tension between the oil and water phases can alsobe used to predict the likelihood that phase inversion will occur in an emulsion (Shinoda andFriberg 1986, Lehnert et al. 1994).

7.9. CHEMICAL AND BIOCHEMICAL STABILITY

The majority of this chapter has been concerned with the physical instability of food emul-sions, rather than with their chemical instability. This is largely because emulsion scientistshave historically focused mainly on the physical aspects of food emulsions. Nevertheless,there are many types of chemical or biochemical reactions which can have adverse effects onthe quality of food emulsions (Fennema 1996a). For this reason, there has been a growinginterest in the influence of various chemical and biochemical reactions on the stability of foodemulsions in recent years.

One of the most common forms of instability in foods which contain fats is lipid oxidation(Nawar 1996). Lipid oxidation leads to the development of undesirable “off-flavors” (rancid-ity) and potentially toxic reaction products. In addition, it may also promote the physicalinstability of some emulsions (Coupland and McClements 1996). For example, many of thereaction products generated during lipid oxidation are surface active and may therefore beable to interact with the interfacial membrane surrounding the droplets in such a way as tolead to droplet coalescence. The importance of lipid oxidation in food emulsions has led toa considerable amount of research being carried out in this area over the past few years(Frankel 1991; Roozen et al. 1994a,b; Coupland and McClements 1996; Coupland et al.1996; Huang et al. 1994, 1996a,b,c; Frankel et al. 1994; Frankel 1996; Mei et al. 1998). Themain emphasis of this work is to develop effective strategies for retarding lipid oxidation in


emulsions by incorporating antioxidants, controlling storage conditions, or engineering drop-let interfacial properties.

There is also an increasing interest in the influence of biochemical reactions on theproperties of food emulsions (Dalgleish 1996a). A number of studies have recently beencarried out to determine the influence of certain enzymes on the stability and physicochemicalproperties of food emulsions (Agboola and Dalgleish 1996b,c). These studies have shownthat the properties of food emulsions can be altered appreciably when the adsorbed proteinsare cleaved by enzyme hydrolysis. A number of studies have also shown that globularproteins become denatured after they have been adsorbed to the surface of an emulsiondroplet and that they remain in this state after they are desorbed (Corredig and Dalgleish1995, de Roos and Walstra 1996). This has important implications for the action of manyenzymes in food emulsions. The activity of an enzyme may be completely lost when it isadsorbed to the surface of the droplets in an emulsion, and therefore any biochemicalreactions catalyzed by it will cease (de Roos and Walstra 1996).

Given their obvious importance for food quality, it seems likely that there will be anincreasing emphasis on the influence of biochemical and chemical reactions on the stabilityof food emulsions in the future.


8 Emulsion Rheology

8.1. INTRODUCTION

The application of a force to a material causes it to become deformed and/or to flow(Whorlow 1992, Macosko 1994). The extent of the deformation and flow depends on thephysicochemical properties of the material. Rheology is the science which is concerned withthe relationship between applied forces and the deformation and flow of matter (Macosko1994). Most rheological tests involve the application of a force to a material and a measure-ment of the resulting flow or change in shape (Whorlow 1992).

A knowledge of the rheological properties of emulsions is important to food scientistsfor a number of reasons (Sherman 1970, Dickinson and Stainsby 1982, Race 1991, Shoe-maker et al. 1992, Rao et al. 1995, Rao 1995). Many of the sensory attributes of foodemulsions are directly related to their rheological properties (e.g., creaminess, thickness,smoothness, spreadability, pourabilty, flowability, brittleness, and hardness). A food manu-facturer must therefore be able to design and produce a product which has the rheologicalproperties expected by the consumer. The shelf life of many food emulsions depends onthe rheological characteristics of the component phases (e.g., the creaming of oil dropletsdepends on the viscosity of the aqueous phase) (Section 7.3). Information about the rheol-ogy of food products is used by food engineers to design processing operations whichdepend on the way that a food behaves when it flows through a pipe, is stirred, or ispacked into containers. Rheological measurements are also used by food scientists as ananalytical tool to provide fundamental insights about the structural organization and inter-actions of the components within emulsions (e.g., measurements of viscosity versus shearrate can be used to provide information about the strength of the colloidal interactionsbetween droplets) (Hunter 1993, Tadros 1994). In this chapter, the basic principles ofrheology, the rheological characteristics of food emulsions, and the instruments available forcarrying out rheological measurements are covered.

Food emulsions are compositionally and structurally complex materials which can exhibita wide range of different rheological behaviors, ranging from low-viscosity fluids (such asmilk and fruit juice beverages) to fairly hard solids (such as refrigerated margarine orbutter). Food scientists aim to develop theories which can be used to describe and predictthe rheological behavior of food emulsions and experimental techniques to characterizethese properties. Despite the diversity of food emulsions, it is possible to characterize manyof their rheological properties in terms of a few simple models: the ideal solid, the idealliquid, and the ideal plastic (Sherman 1970, Tung and Paulson 1995). More complexsystems can then be described by combining two or more of these simple models. In thefollowing sections, the concepts of the ideal solid, ideal liquid, and ideal plastic are intro-duced, as well as some of the deviations from these models that are commonly observed infood emulsions.


8.2. RHEOLOGICAL PROPERTIES OF MATERIALS

8.2.1. Solids

In our everyday lives, we come across solid materials which exhibit quite different rheologi-cal characteristics. Some may be soft, others hard; some may be brittle, others rubbery; somemay break easily, others may not. Despite this range of different behaviors, it is possible tocharacterize the rheological properties of many solid foods in terms of a few simple concepts.

8.2.1.1. Ideal Elastic Solids

An ideal elastic solid is often referred to as a Hookean solid, after Robert Hooke, the scientistwho first described this type of behavior (Whorlow 1992, Macosko 1994, Rao et al. 1995).Hooke observed experimentally that there is a linear relationship between the deformation ofa solid material and the magnitude of the force applied to it, provided the deformation is nottoo large (Figure 8.1). He also observed that when the force was removed from the material,it returned back to its original length. In general, Hooke found that the force per unit area (orstress) was proportional to the relative deformation (or strain). Hooke’s law can therefore besummarized by the following statement:

stress constant strain ( ) ( ) ( )τ γ= ×E (8.1)

A stress can be applied to a material in a number of different ways, including simple shear,simple compression, and bulk compression (Figure 8.2). Equation 8.1 is applicable to eachof these situations, but the values of the stress, strain, and constant used depend on the natureof the deformation (Table 8.1).

The equations given in Table 8.1 assume that the material is homogeneous and isotropic(i.e., its properties are the same in all directions). To characterize the rheological constantsof an ideal elastic solid, it is therefore necessary to measure the change in its dimensions whena force of known magnitude is applied.

The elastic behavior of a solid is related to the intermolecular forces which hold themolecules together. When a stress is applied to a material, the bonds between the moleculesare compressed or expanded, and therefore they store energy. When the stress is removed, thebonds give up this energy and the material returns to its original shape. The elastic modulus

FIGURE 8.1 At small deformations, there is a linear relationship between the applied stress and theresultant strain for an ideal elastic solid. At higher deformations, the stress is no longer linearly relatedto strain and the material will eventually break.


FIGURE 8.2 An elastic solid can be deformed in a number of different ways.

of an ideal elastic solid is therefore related to the strength of the interactions between themolecules within it.

8.2.1.2. Nonideal Elastic Solids

Hooke’s law is only strictly applicable to elastic materials at low strains, and therefore mostfundamental rheological studies of solid foods are carried out using very small materialdeformations (<1%). Nevertheless, the rheological behavior of foods at large deformations isoften more relevant to their actual use (e.g., spreadability, slicing, or mastication) (van Vliet1995). For this reason, it is also important to characterize the rheological behavior of solidsat large deformations. At strains just above the region where Hooke’s law is obeyed, the stressis no longer proportional to the strain, and therefore an apparent modulus is defined, whichis equal to the stress/strain at a particular value of the strain. It is therefore necessary tostipulate the strain (or stress) at which an apparent modulus of a material is measured. Eventhough the material does not strictly obey Hooke’s law, it still returns to its original shapeonce the force is removed. Above a certain deformation, however, a solid may not return toits original shape after the applied stress is removed, because it either breaks or flows. A

TABLE 8.1Rheological Parameters for Different Types of Deformation of Elastic Solids

Deformation Stress Strain Elastic modulus

Simple shear τ γ φφ

= = = =FA

ll

GF

A∆

tantan

Simple compression τ γ= = =FA

ll

YFl

A l∆

∆

Bulk compression τ γ= = = =FA

PV

VK

PVV

∆∆

Note: G is the shear modulus, Y is Young’s modulus, K is the bulk modulus, P is the pressure, φ is thedeformation angle, and the other symbols are as defined in Figure 8.2.


material that breaks at low strains is referred to as being brittle, whereas a material that flowsis referred to as being plastic or viscoelastic (see later). The stress at which a material breaksis referred to as the breaking stress, whereas the strain at which it breaks is referred to as thebreaking strain. A material usually ruptures or flows when the forces holding the atoms ormolecules in the material together are exceeded. This often begins at regions where the bondsholding the material together are relatively weak (e.g., a crack). A knowledge of the breakingstress or breaking strain required to disrupt a material is often a useful indication of its abilityto be broken up during mastication, cut with a knife, or its sensitivity to rupture during storageand transport.

8.2.2. Liquids

Liquid food emulsions also exhibit a wide range of rheological properties. Some have lowviscosities and flow easily, like milk, while others are very viscous, like mayonnaise. Evenso, it is possible to characterize their rheological properties using a few simple concepts.

8.2.2.1. Ideal Liquids

The ideal liquid is often referred to as a Newtonian liquid, after Isaac Newton, the scientistwho first described its behavior (Whorlow 1992, Macosko 1994, Rao 1995). When a shearstress is applied to an ideal liquid, it continues to flow as long as the stress is applied. Oncethe stress is removed, there is no elastic recovery of the material (i.e., it does not return toits original shape).

The viscosity of a liquid is a measure of its resistance to flow: the higher the viscosity, thegreater the resistance (Macosko 1994). The concept of viscosity can be understood byconsidering a liquid which is contained between two parallel plates (Figure 8.3). The bottomplate is at rest, while the top plate moves in the x-direction with a constant velocity (v). It isassumed that the liquid between the plates consists of a series of infinitesimally thin layers.The liquid layers in direct contact with the bottom and top plates are assumed to “stick” tothem, so that they have velocities of 0 and v, respectively. The intervening liquid layers slideover each other with velocities that range between 0 and v, the actual value being given bydy(dv/dy), where dy is the distance from the bottom plate and dv/dy is the velocity gradientbetween the plates. The shear stress applied to the fluid is equal to the shear force dividedby the area over which it acts (τ = F/A). The rate of strain is given by the change indisplacement of the layers per unit time: dγ/dt (or «γ) = dv/dy. For an ideal liquid, the shearstress is proportional to the rate of strain (Figure 8.4):

τ ηγ= « (8.2)

FIGURE 8.3 The viscosity of a liquid is related to the friction between the liquid layers as they slideacross each other: the greater the friction, the higher the viscosity.


where the constant of proportionality (η) is called the viscosity. The viscosity arises from thefriction between the liquid layers as they slide past one another (Macosko 1994). The lowerthe viscosity of a liquid, the less resistance between the liquid layers, and therefore thesmaller the force required to cause the top plate to move with a given velocity, or the fasterthe top plate moves when a given force is applied. The ideal viscous fluid differs from theideal elastic solid because the shear stress is proportional to the rate of strain (Figure 8.3),rather than the strain (Figure 8.1).

The units of shear stress (τ) are Ν m–2 (or Pa), and those of shear rate ( «γ) are s–1; thus theviscosity (η) has units of N s m–2 (or Pa s) in the SI system. Viscosity can also be expressedin the older cgs units of poise, where 1 Pa s = 10 P. Thus the viscosity of water can be quotedas 1 mPa s, 0.001 Pa s, 0.01 P, or 1 cP, depending on the units used.

Ideally, a Newtonian liquid should be incompressible (its volume does not change whena force is applied to it), isotropic (its properties are the same in all directions), and structure-less (it is homogeneous). Although many liquid foods do not strictly meet these criteria, theirrheological behavior can still be described excellently by Equation 8.2 (e.g., milk). Neverthe-less, there are many others that exhibit nonideal liquid behavior and so their properties cannotbe described by Equation 8.2.

The type of flow depicted in Figure 8.3 occurs at low shear rates and is known as laminarflow, because the liquid travels in a well-defined laminar pattern. At higher shear rates, eddiesform in the liquid and the flow pattern is much more complex. This type of flow is referredto as turbulent, and it is much more difficult to mathematically relate the shear stress to therate of strain under these conditions. For this reason, instruments that measure the viscosityof liquids are designed to avoid turbulent flow.

8.2.2.2. Nonideal Liquids

Nonideality may manifest itself in a number of different ways; for example, the viscosity ofa liquid may depend on the shear rate and/or the time over which the shear stress is applied,or the fluid may exhibit some elastic as well as viscous properties (Macosko 1994, Tung andPaulson 1995). Plastic and viscoelastic materials, which have some elastic characteristics, areconsidered in later sections.

Shear-Rate-Dependent Nonideal Liquids. In an ideal liquid, the viscosity is independent ofshear rate and the length of time the liquid is sheared (i.e., the ratio of the shear stress to theshear rate does not depend on shear rate or time) (Figure 8.5). In practice, many foodemulsions have viscosities which depend on the shear rate and the length of time the emulsionis sheared (Dickinson 1992). In this section, emulsions in which the viscosity depends onshear rate but is independent of the shearing time are examined (Dickinson 1992). In thefollowing section, emulsions in which the viscosity depends on both the shear rate andshearing time are examined.

FIGURE 8.4 Stress is proportional to rate of strain for an ideal liquid.


(a) (b)

The viscosity of an emulsion may either increase or decrease as the shear rate is increased,rather than staying constant, as for a Newtonian liquid (Figure 8.5). In these systems, theviscosity at a particular shear rate is referred to as the apparent viscosity. The dependenceof the apparent viscosity on shear rate means that it is crucial to stipulate the shear rate usedto carry out the measurements when reporting data. The choice of shear rate when measuringthe apparent viscosity of a nonideal liquid is a particularly important consideration whencarrying out rheological measurements in a laboratory which are supposed to mimic someprocess which occurs in a food naturally (e.g., flow through a pipe, creaming of an emulsiondroplet, or mastication). The test in the laboratory should use a shear rate that is as close aspossible to that which the food experiences in practice.

The two most common types of shear-rate-dependent nonideal liquids are:

1. Pseudoplastic fluids. Pseudoplastic flow is the most common type of nonidealbehavior exhibited by food emulsions. It manifests itself as a decrease in theapparent viscosity of a fluid as the shear rate is increased and is therefore oftenreferred to as shear thinning (Figure 8.5). Pseudoplasticity may occur for a varietyof reasons in food emulsions (e.g., the spatial distribution of the particles may bealtered by the shear field, nonspherical particles may become aligned with the flowfield, solvent molecules bound to the particles may be removed, or flocs may bedeformed and disrupted) (Hunter 1993, Mewis and Macosko 1994).

2. Dilatant fluids. Dilatant behavior is much less common than pseudoplastic be-havior. It manifests itself as an increase in the apparent viscosity as the shear rateis increased and is therefore often referred to as shear thickening (Figure 8.5).Dilantancy is often observed in concentrated emulsions or suspensions where theparticles are packed tightly together (Hunter 1989). At intermediate shear rates, theparticles form two-dimensional “sheets” which slide over each other relativelyeasily, but at higher shear rates, these sheets are disrupted and so the viscosityincreases (Pal 1996). Shear thickening may also occur when the particles in anemulsion become flocculated because of an increased collision frequency (Section7.4.1.1); therefore, this process usually leads to time-dependent behavior and sowill be considered in the following section.

Liquids that exhibit pseudoplastic behavior often have a viscosity versus shear rate profilesimilar to that shown in Figure 8.6. The viscosity decreases from a constant value at low shearrates (η0) to another constant value at high shear rates (η∞). A number of mathematicalequations have been developed to describe the rheological behavior of shear-rate-dependentnonideal liquids. The major difference is the range of shear rates over which they are

FIGURE 8.5 Comparison of the viscosity of ideal and nonideal liquids. (a) Shear stress versus shearrate. (b) Viscosity versus shear rate.


FIGURE 8.6 Typical viscosity profile for a pseudoplastic material. The viscosity decreases from aconstant value (η0) at low shear rates to another constant value (η∞) at high shear rates. (Adapted fromHunter 1993.)

applicable. If measurements are carried out across the whole shear rate range (which oftenextends many orders of magnitude), then the viscosity can often be described by the Meterequation (Hunter 1989):

η ηη η

τ τ= +

−+∞

∞0

1 ( / )in (8.3)

where τ i is the shear stress where the viscosity is midway between the low and high shear ratelimits and n is the power index. The rheological properties of this type of system can thereforebe characterized by four parameters: η0, η∞, τi , and n.

If measurements are only carried out at shear rates that are sufficiently less than the highshear rate plateau (Figure 8.6), then the rheology can be described by the Ellis equation(Hunter 1993):

ηητ τ

=+

0

1 ( / )in (8.4)

If measurements are carried out at intermediate shear rates (i.e., above the low shearplateau and below the high shear plateau), the rheology can often be described by a simplepower-law model (Hunter 1993):

τ γ η γ= = −A d dt A d dtB B( / ) ( / )or 1 (8.5)

The constants A and B are usually referred to as the consistency index and the power index,respectively (Dickinson 1992). For an ideal liquid, B = 1; for an emulsion which exhibitsshear thinning, B < 1; and for an emulsion which exhibits shear thickening, B > 1. Equations8.5 are easy to use since they only contain two unknown parameters, which can simply beobtained from a plot of log (τ) versus log (dγ/dt). Nevertheless, these equations should onlybe used after it has been proven experimentally that the relationship between log (τ) and log(dγ/dt) is linear over the shear rates used.

Time-Dependent Nonideal Liquids. The apparent viscosity of the fluids described in theprevious section depended on the shear rate, but not on the length of time that the shear was


applied. There are many food emulsions whose apparent viscosity either increases or de-creases with time during the application of shear. In some cases, this change is reversible andthe fluid will recover its original rheological characteristics if it is allowed to stand at rest fora sufficiently long period. In other cases, the change brought about by shearing the sampleis irreversible, and the sample will not recover its original characteristics.

An appreciation of the time dependency of the flow properties of food emulsions is of greatpractical importance in the food industry. The duration of pumping or mixing operations, forinstance, must be carefully controlled so that the food sample has an apparent viscosity whichis suitable for the next processing operation. If a food is mixed or pumped for too long, it maybecome too thick or too runny and thus lose its desirable rheological properties.

The dependence of the rheology of a liquid on time is often associated with some kind ofrelaxation process (Hunter 1993, Mewis and Macosko 1994). When an external force isapplied to a system that is initially at equilibrium, the material takes a certain length of timeto reach the new equilibrium conditions, which is characterized by a relaxation time (τR).When the measurement time is of the same order as the relaxation time, it is possible toobserve changes in the properties of the system with time. Thus the rheological properties ofan emulsion depend on the time scale of the experiment. Time-dependent nonideal fluids areclassified in two different categories:

1. Thixotropic behavior. A thixotropic fluid is one in which the apparent viscositydecreases with time when the fluid is subjected to a constant shear rate (Figure8.7). Emulsions which exhibit this type of behavior often contain particles (drop-lets, crystals, or biopolymers) which are aggregated by weak forces. Shearing ofthe material causes the aggregated particles to be progressively deformed anddisrupted, which decreases the resistance to flow and therefore causes a reductionin viscosity over time. If the relaxation time associated with the disruption of theflocs is shorter than the measurement time, then the viscosity will be observed totend to a constant final value. This value may correspond to the point where therate of structure disruption is equal to the rate of structure reformation or wherethere is no more structure to be broken down. In pseudoplastic liquids, the break-down of the aggregated particles occurs so rapidly that the system almost imme-diately attains its new equilibrium position, and so it appears as though the viscos-ity is independent of time.

2. Rheopectic. In some food emulsions, the apparent viscosity of the fluid increaseswith time when it is subjected to a constant shear rate (Figure 8.7). One of the mostcommon reasons for this type of behavior is that shearing increases both thefrequency and efficiency of collisions between droplets, which leads to enhancedaggregation (Section 7.4.1) and consequently an increase in apparent viscosityover time.

In some fluids, the time-dependent rheological properties are irreversible (i.e., once theshear force is removed, the system may not fully regain its initial rheological properties).Liquids that experience this type of permanent change are called rheodestructive. This typeof behavior might occur when flocs are disrupted by an intense shear stress and are unableto reform when the shear stress is removed. Otherwise, the structure and rheological prop-erties of a material may return to their original values. In this case, the recovery time is oftenan important characteristic of the material.

The rheological properties of time-dependent nonideal liquids can be characterized bymeasuring the change in their viscosity over time. From these measurements, one can obtaina relaxation time for the structural rearrangements that occur in the emulsion. Nevertheless,


FIGURE 8.8 Hysteresis curve for liquids whose viscosity depends on the length of time they aresheared.

FIGURE 8.7 Comparison of the viscosity of ideal and time-dependent nonideal liquids. The viscositymay increase or decrease to a constant value with time. Alternatively, it may increase steeply due tonetwork formation.

this type of experiment is often inconvenient if one also wants to obtain information aboutthe dependence of the viscosity on shear rate. One would have to establish the relaxation timefor the structural rearrangements at each shear rate, and then ensure that the samples weresheared for a time that was long enough for them to reach their steady-state rheology. If therelaxation time of a sample is relatively long, this type of measurement would be timeconsuming and laborious to carry out. Instead, it is often more convenient to measure theviscosity of a fluid when the shear rate is increased from zero to a certain value and thendecreased back to zero again (Figure 8.8). When there is a significant structural relaxation ina system, the upward curve is different from the downward curve and one obtains a hysteresisloop. The area within the loop depends on the degree of relaxation that occurs and the rateat which the shear rate is altered. The slower the shear rate is altered, the more time the systemhas to reach its equilibrium value, and therefore the smaller the area within the hysteresisloop. By carrying out measurements as a function of the rate at which the shear rate isincreased, it is possible to obtain information about the relaxation time.

8.2.3. Plastics

A number of food emulsions exhibit rheological behavior known as plasticity (e.g., marga-rine, butter, and certain spreads) (Sherman 1968a,c, 1970; Tung and Paulson 1995). A plasticmaterial has elastic properties below a certain applied stress, known as the yield stress, butflows like a fluid when this stress is exceeded.


FIGURE 8.9 Rheological behavior of ideal and nonideal plastics.

8.2.3.1. Ideal Plastics

The ideal plastic material is referred to as a Bingham plastic, after the scientist who firstproposed this type of rheological behavior (Sherman 1970). Two equations are needed todescribe the rheological behavior of a Bingham plastic, one below the yield stress and oneabove it:

τ γ τ τ= <G ( )for 0 (8.6)

τ τ ηγ τ τ− = ≥0 0« ( )for (8.7)

where G is the shear modulus, η is the viscosity, and τ0 is the yield stress. The rheologicalproperties of an ideal plastic are shown in Figure 8.9.

Foods that exhibit plastic behavior usually consist of a network of aggregated moleculesor particles dispersed in a liquid matrix (Clark 1987, Edwards et al. 1987, Tung and Paulson1995). For example, margarine and butter consist of a network of tiny fat crystals dispersedin a liquid oil phase (Moran 1994). Below a certain applied stress, there is a small deforma-tion of the sample, but the weak bonds between the crystals are not disrupted. When thecritical yield stress is exceeded, the weak bonds are broken and the crystals slide past oneanother, leading to flow of the sample. Once the force is removed, the flow stops. A similartype of behavior can be observed in emulsions containing three-dimensional networks ofaggregated droplets.

8.2.3.2. Nonideal Plastics

Above the yield stress, the fluid flow may exhibit non-Newtonian behavior similar to thatdescribed earlier for liquids (e.g., pseudoplastic, dilatant, thixotropic, or rheopectic). Thematerial may also exhibit nonideal elastic behavior below the yield stress (e.g., the yield pointmay not be sharply defined; instead, the stress may increase dramatically, but not instanta-neously, as the shear rate is increased) (Figure 8.9). This would occur if the material did notall begin to flow at a particular stress, but there was a gradual breakdown of the networkstructure over a range of stresses (Sherman 1968a,c).

8.2.4. Viscoelastic Materials

Many food emulsions are not pure liquids or pure solids, but have rheological properties thatare partly viscous and partly elastic (Sherman 1968a,c, 1970; Dickinson 1992). Plasticmaterials exhibit elastic behavior below a certain value of the applied stress and viscous


behavior above this value. In contrast, viscoelastic materials exhibit both viscous and elasticbehavior simultaneously. In an ideal elastic solid, all the mechanical energy applied to thematerial is stored in the deformed bonds and is returned to mechanical energy once the forceis removed (i.e., there is no loss of mechanical energy). On the other hand, in an ideal liquid,all of the mechanical energy applied to the material is dissipated due to friction (i.e., themechanical energy is converted to heat). In a viscoelastic material, part of the energy is storedas mechanical energy within the material, and part of the energy is dissipated. For this reason,when a force is applied to a viscoelastic material, it does not instantaneously adopt its newdimensions, nor does it instantaneously return to its undeformed state when the force isremoved (as an ideal elastic material would). In addition, the material may even remainpermanently deformed once the force is removed. The rheological properties of a viscoelasticmaterial are characterized by a complex elastic modulus (E*) which is comprised of an elasticand a viscous contribution:

E E iE* = ′ + ′′ (8.8)

Here, E′ is known as the storage modulus and E″ as the loss modulus.Two types of experimental tests are commonly used to characterize the rheological prop-

erties of viscoelastic materials: one based on transient measurements and the other ondynamic measurements (Whorlow 1992). Both types of tests can be carried out by theapplication of simple shear, simple compression, or bulk compression to the material beinganalyzed. Simple shear tests are the most commonly used to analyze food emulsions, andtherefore only these will be considered here. Nevertheless, the same basic principles are alsorelevant to compression tests.

8.2.4.1. Transient Tests

In a transient experiment, a constant stress is applied to a material and the resulting strain ismeasured as a function of time or vice versa.

Creep. In a creep experiment, a constant stress is applied to a material and the change in itsdimensions with time are monitored, which results in a strain versus time curve (Sherman1968c, 1970). The data are usually expressed in terms of a parameter called the compliance(J), which is equal to the ratio of the strain to the applied stress (and is therefore the reciprocalof the modulus). The compliance is proportional to the strain, but it is a better parameter touse to characterize the rheological properties of the material because it takes into account themagnitude of the applied stress. The time dependence of the compliance of a material can alsobe measured when the stress is removed, which is referred to as a creep recovery experiment.A typical compliance versus time curve for a viscoelastic material is shown in Figure 8.10(Sherman 1968a). This curve can be divided into three regions:

1. A region of instantaneous elastic deformation in which the bonds between theparticles are stretched elastically. In this region, the material acts like an elasticsolid with a compliance J0 given by the ratio of the strain to the applied stress.

2. A region of retarded elastic compliance in which some bonds are breaking andsome are reforming. In this region, the material has viscoelastic properties and itscompliance is given by JR = JM [1 – exp(–t/τM)], where JM and τM are the meancompliance and retardation time.

3. A region of Newtonian compliance (JN) when the bonds are disrupted and do notreform so that the material only flows: JN = t/ηN.


FIGURE 8.10 A typical creep versus time curve for a viscoelastic material, such as ice cream.

The total creep compliance of the system is therefore given by:

J t J J t J t J J t tR N M M N( ) ( ) ( ) [ exp( / )] /= + + = + − − +0 0 1 τ η (8.9)

This type of material is usually referred to as a viscoelastic liquid, because it continues toflow for as long as the stress is applied. Some materials exhibit a different type of behaviorand are referred to as viscoelastic solids. When a constant stress is applied to a viscoelasticsolid, the creep compliance increases up to a finite equilibrium value (JE) at long times ratherthan continuously increasing. When the force is removed, the compliance returns to zero,unlike a viscoelastic liquid, which does not return to its initial shape.

Stress Relaxation. Instead of applying a constant force and measuring the change in thestrain with time, it is also possible to apply a constant strain and measure the change in thestress acting on the material with time. This type of experiment is referred to as a stressrelaxation. The same type of information can be obtained from creep and stress relaxationexperiments, and the method used largely depends on the type of rheological instrumentavailable.

8.2.4.2. Dynamic Tests

In a dynamic experiment, a sinusoidal stress is applied to a material and the resultingsinusoidal strain is measured or vice versa (Tung and Paulson 1995, Liu and Masliyah 1996).In this section, only the case where a stress is applied to the sample and the resultant strainis measured is considered. The applied stress is characterized by its maximum amplitude (τ0)and its angular frequency (ω). The resulting strain has the same frequency as the appliedstress, but its phase is different because of relaxation mechanisms associated with the material(Whorlow 1992). Information about the viscoelastic properties of the material can thereforebe obtained by measuring the maximum amplitude (γ0) and phase shift (δ) of the strain(Figure 8.11). The amplitude of the applied stress used in this type of test is usually so smallthat the material is in the linear viscoelastic region (i.e., the stress is proportional to thestrain), and the properties of the material are not affected by the experiment (van Vliet 1995,Liu and Masliyah 1996).

If the applied stress varies sinusoidally with time, then (Whorlow 1992):

τ τ ω= 0 cos( )t (8.10)

and the resulting harmonic strain is


FIGURE 8.11 The rheological properties of a viscoelastic material can be determined by measuringthe relationship between an applied sinusoidal stress and the resultant sinusoidal strain.

γ γ ω δ= −0 cos( )t (8.11)

The compliance of the material is therefore given by:

J t t t( ) (cos cos sin sin )= = +γτ

γτ

δ ω δ ω0

0

0(8.12)

or

J t J t J t( ) cos sin= ′ + ′′ω ω (8.13)

where J′ (= γ0 cos δ/τ0) is known as the storage compliance, which is the in-phase componentof the compliance, and J″ (= γ0 sin δ/τ0) is known as the loss compliance, which is the 90°out-of-phase component of the compliance. The in-phase component of the compliance isdetermined by the elastic properties of the material, whereas the 90° out-of-phase componentis determined by the viscous properties. This is because the stress is proportional to the strain(τ ∝ γ) for elastic materials, whereas it is proportional to the rate of strain (τ ∝ dγ/dt) forviscous materials (Macosko 1994).

The dynamic rheological properties of a material can therefore be characterized by mea-suring the frequency dependence of the applied stress and the resulting strain and thenplotting a graph of J′ and J″ versus frequency. Alternatively, the data are often presented interms of the magnitude of the complex compliance (J* = J ′ – iJ″) and the phase angle:

J J J* = ′ + ′′2 2 (8.14)

δ = ′′′

−tan 1 JJ

(8.15)


The phase angle of a material provides a useful insight into its viscoelastic properties: δ= 0° for a perfectly elastic solid, δ = 90° for a perfectly viscous fluid, and 0 < δ < 90° fora viscoelastic material. The more elastic a material (at a particular frequency), the smaller thephase angle and the lower the amount of energy dissipated per cycle.

It is often more convenient to express the rheological properties of a viscoelastic materialin terms of its modulus rather than its compliance (Whorlow 1992). The complex, storage,and loss moduli of a material can be calculated from the measured compliances using thefollowing relationships:

G G iG GJ

J JG

JJ J

* = ′ + ′′ ′ = ′′ + ′′

′′ = ′′′ + ′′2 2 2 2 (8.16)

8.3. MEASUREMENT OF RHEOLOGICAL PROPERTIES

Food emulsions can exhibit a wide range of different types of rheological behavior, includingliquid, solid, plastic, and viscoelastic (Dickinson and Stainsby 1982, Dickinson 1992). Con-sequently, a variety of instrumental methods have been developed to characterize theirrheological properties. Instruments vary according to the type of deformation they apply tothe sample (shear, compression, elongation, or some combination), the property they mea-sure, their cost, their sophistication, and their ease of operation (Whorlow 1992).

In many industrial applications, it is necessary to have instruments that make measure-ments which are rapid, low cost, simple to carry out, and reproducible, rather than giveabsolute fundamental data (Sherman 1970, Rao 1995). Thus simple empirical instruments areoften used in quality assurance laboratories, rather than the more sophisticated and expensiveinstruments used in research and development. The information obtained from these empiri-cal instruments is often difficult to relate to the fundamental rheological constants of amaterial because the applied stresses and strains are not easily measured or defined. Ratherthan a simple elongation, shear, or compression, different types of forces may be appliedsimultaneously. For example, when a blade cuts through a meat product, both shear andcompression forces are applied together, and the sample is deformed beyond the limit whereHooke’s law is applicable. To compare data from different laboratories, it is necessary tocarefully follow standardized test procedures. These procedures may define experimentalparameters such as the sample size and preparation procedure, the magnitude of the force ordeformation, the design of the device used, the speed of the probe, the length of time the forceis applied, and the measurement temperature.

For food scientists involved in research and development, it is usually necessary to useinstruments which provide information about the fundamental rheological constants ofthe material being tested. These instruments are designed to apply well-defined stressesand strains to a material in a controlled manner so that stress–strain relationships can bemeasured and interpreted using available mathematical theories. Rheological propertiesdetermined using these techniques can be compared with measurements made by otherworkers or in other laboratories. In addition, measured rheological properties can becompared with predictions made using various mathematical theories which have beendeveloped to relate the structure and composition of materials to their fundamental rheologi-cal properties.

It is convenient to categorize rheological instruments according to whether they utilizesimple compression (or elongation) or shear forces.*

* At present, few instruments utilize bulk compression to analyze the rheological properties of food emulsions.


FIGURE 8.12 Universal Testing Machine for measuring the rheological properties of a material bya compression or elongation test.

8.3.1. Simple Compression and Elongation

This type of test is most frequently carried out on solid or semisolid foods that are capableof supporting their own weight (e.g., butter, margarine, and frozen ice cream) (Bourne 1982,Rao et al. 1995). Measurements are often carried out using instruments referred to as Uni-versal Testing Machines. The solid sample to be analyzed is placed between a fixed plate anda moving probe (Figure 8.12). The probe can have many different designs depending on thetype of information required, including a flat plate, a blade, a cylindrical spike, and even aset of teeth!

The probe can be moved vertically, either upward or downward, at a controlled speed.Either the probe or the plate contains a pressure sensor which measures the force exerted onthe sample when it is deformed by the probe. The instrument also records the distance thatthe probe moves through the sample. The stress and strain experienced by a material cantherefore be calculated from a knowledge of its dimensions and the force and deformationrecorded by the instrument.

Some of the common tests carried out using Universal Testing Machines are:

1. Stress–strain curve. The stress on a sample is measured as a function of strain asit is compressed at a fixed rate (Figure 8.1). The resulting stress–strain curve isused to characterize the rheological properties of the material being tested. Theslope of stress versus strain at small deformations is often a straight line, with agradient equal to the elastic modulus (Table 8.1). At intermediate deformations, thestress may no longer be proportional to the strain and some flow may occur, so thatwhen the stress is removed the sample does not return to its original shape. Atlarger deformations, the sample may rupture and the breaking stress, strain, andmodulus can be determined. The operator must decide the distance and speed atwhich the probe will move through the sample. For viscoelastic materials, theshape of the upward and downward curves may be different and depends on thespeed at which the probe moves. This type of test is used commonly to test solidsamples and gels, such as margarine, butters, spreads, and desserts.

2. Repeated deformation. The sample to be analyzed is placed between the plateand the probe, and then the probe is lowered and raised a number of times at a fixedspeed so that the sample experiences a number of compression cycles (Rao et al.1995). An ideal elastic solid would show the same stress–strain curve for eachcycle. However, the properties of many materials are altered by compression (e.g.,due to rupture or flow), and therefore successive compression cycles give differentstress–strain curves. This type of test is often used to give some indication of the


processes that occur when a food is chewed in the mouth (i.e., the breakdown offood structure).

3. Transient experiments. The sample is placed between the plate and the probe andthen compressed to a known deformation, and the relaxation of the stress with timeis measured (stress relaxation). Alternatively, a constant stress could be applied tothe sample, and the variation of the strain is measured over time (creep). This typeof experiment is particularly useful for characterizing the rheological properties ofviscoelastic food emulsions (see Section 8.2.4).

By using different fixtures, the same type of instrument can be used to carry out elongationexperiments. A sample is clamped at both ends, and then the upper clamp is moved upwardat a controlled speed and the force required to elongate the sample is measured by thepressure sensor as a function of sample deformation. Again, the elastic modulus and breakingstrength of the material can be determined by analyzing the resulting stress–strain relation-ship. Universal Testing Machines can also be adapted to perform various other types ofexperiments, such as bending or slicing.

Recently, a number of more sophisticated instruments, based on dynamic rheologicalmeasurements, have been developed to characterize the rheological properties of solids,plastics, and viscoelastic materials (Harwalker and Ma 1990, Wunderlich 1990, Whorlow1992). In addition to carrying out the standard compression measurements mentioned above,they can also be used to carry out dynamic compression measurements on viscoelasticmaterials. The sample to be analyzed is placed between a plate and a probe, and an oscillatoryshear stress of known amplitude and frequency is applied to it. The amplitude and phase ofthe resulting strain are measured and converted into a storage and loss modulus using suitableequations (Section 8.2.4.2). The amplitude of the applied stress must be small enough to bein the linear viscoelastic region of the material. These instruments are relatively expensive topurchase and therefore only tend to be used by research laboratories in large food companies,government institutions, and universities. Nevertheless, they are extremely powerful tools forcarrying out fundamental studies of food emulsions. The rheological properties of a samplecan be measured as a function of time or temperature, and thus processes such as gelation,aggregation, crystallization, melting, and glass transitions can be monitored. The measure-ment frequency can also be varied, which provides valuable information about relaxationprocesses within a sample.

Some complications can arise when carrying out simple compression experiments. Theremay be friction between the compressing plates and the sample, which can lead to thegeneration of shear as well as compressional forces (Whorlow 1992). For this reason, it isoften necessary to lubricate the sample with oil to reduce the effects of friction. In addition,the cross-sectional area of the sample may change during the course of the experiment, whichwould have to be taken into account when converting the measured forces into stresses.Finally, for viscoelastic materials, some stress relaxation may occur during the compressionor expansion, and so the results depend on the rate of sample deformation.

8.3.2. Shear Measurements

Instruments which utilize shear measurements are used to characterize the rheological prop-erties of liquids, viscoelastic materials, plastics, and solids (Whorlow 1992, Rao 1995). Thetype of instrument and test method used in a particular situation depend on physicochemi-cal characteristics of the sample being analyzed, as well as the kind of information required.Some instruments can be used to characterize the rheological properties of both solids andliquids, whereas others can only be used for either solids or liquids. Certain types of viscom-


eters are capable of measuring the viscosity of fluids over a wide range of shear rates and cantherefore be used to analyze both ideal and nonideal liquids, whereas the shear rate cannotbe controlled in others and so they are only suitable for analyzing ideal liquids. A number ofinstruments can be used to characterize the rheological behavior of viscoelastic materialsusing both transient and dynamic tests, whereas others can only use either one or the othertype of test. To make accurate and reliable measurements, it is important to select the mostappropriate instrument and test method and to be aware of possible sources of experimentalerror.

8.3.2.1. Capillary Viscometers

The simplest and most commonly used capillary viscometer is called the Ostwald viscometer(Hunter 1986, Whorlow 1992). This device consists of a glass U-tube into which the sampleto be analyzed is poured. The whole arrangement is placed in a thermostated water bath toreach the measurement temperature (Figure 8.13). The viscosity of the liquid is measured bysucking it into one arm of the tube using a slight vacuum and then measuring the time it takesto flow back through a capillary of fixed radius and length. The time it takes to travel throughthe capillary is related to the viscosity by the following equation:

t C= ηρ (8.17)

where ρ is the density of the fluid, t is the measured flow time, and C is a constant whichdepends on the precise size and dimensions of the U-tube. The higher the viscosity of thefluid, the longer it takes to flow through the tube. The simplest method for determining theviscosity of a liquid is to measure its flow time and compare it with that of a liquid of knownviscosity, such as distilled water:

ηρρ

ηss st

t=

0 0

0 (8.18)

FIGURE 8.13 Capillary viscometer used to measure the viscosity of Newtonian liquids.


FIGURE 8.14 Different types of measurement cells commonly used with dynamic shear rheometersand viscometers.

where the subscripts s and 0 refer to the sample being analyzed and the reference fluid,respectively. This type of viscometer is used principally to measure the viscosity of Newtonianliquids. It is unsuitable for analyzing non-Newtonian liquids because the sample does notexperience a uniform and controllable shear rate (Hunter 1986). U-tubes with capillaries ofvarious diameters are available to analyze liquids with different viscosities: the larger thediameter, the higher the viscosity of the sample which can be analyzed.

8.3.2.2. Mechanical Viscometers and Dynamic Rheometers

A number of mechanical rheological instruments have been designed to measure the shearproperties of liquids, viscoelastic materials, plastics, and solids (Whorlow 1992, Macosko1994). These instruments are usually computer controlled and can carry out sophisticated testprocedures as a function of time, temperature, shear rate, or frequency (Figure 8.14). Basi-cally, the sample to be analyzed is placed in a thermostated measurement cell, where it issubjected to a controlled shear stress (or strain). The resulting strain (or stress) is measuredby the instrument, and so the rheological properties of the sample can be determined from thestress–strain relationship. The type of rheological test carried out depends on whether thesample is liquid, solid, or viscoelastic. The instruments can be divided into two differenttypes: constant stress instruments, which apply a constant torque to the sample and measurethe resultant strain or rate of strain, and constant strain instruments, which apply a constantstrain or rate of strain and measure the torque generated in the sample. For convenience, onlyconstant stress instruments are discussed here, although both types are commonly used in thefood industry.

A number of different types of measurement cells can be used to contain the sample duringan experiment (Pal et al. 1992):

1. Concentric cylinder. The sample is placed in the narrow gap between two con-centric cylinders. The inner cylinder is driven at a constant torque (angular force)and the resultant strain (angular deflection) or rate of strain (speed at which thecylinder rotates) is measured, depending on whether one is analyzing a predomi-nantly solid or liquid sample.* For a solid, the angular deflection of the innercylinder from its rest position is an indication of its elasticity: the larger thedeflection, the smaller the shear modulus. For a liquid, the speed at which the innercylinder rotates is governed by the viscosity of the fluid between the plates: the

* In some instruments, the outer cylinder rotates and the torque on the inner cylinder is measured.


faster it spins at a given torque, the lower the viscosity of the liquid being analyzed.The torque can be varied in a controlled manner so that the (apparent) elasticmodulus or viscosity can be measured as a function of shear stress. This instrumentcan be used for measuring the viscosity of non-Newtonian liquids, the viscoelas-ticity of semisolids, and the elasticity of solids.

2. Parallel plate. In this type of measurement cell, the sample is placed between twoparallel plates. The lower plate is stationary, while the upper one can rotate. Aconstant torque is applied to the upper plate, and the resultant strain or rate ofstrain is measured, depending on whether one is analyzing a predominantly solidor liquid sample. The main problem with this type of experimental arrangement isthat the shear strain varies across the sample: the shear strain in the middle of thesample is less than that at the edges. The parallel plate arrangement is thereforeonly suitable for analyzing samples which have rheological properties that areindependent of shear rate, and it is therefore unsuitable for analyzing nonidealliquids or solids.

3. Cone and plate. This is essentially the same design as the parallel plate instru-ment, except that the upper plate is replaced by a cone. The cone has a slight anglewhich is designed to ensure that a more uniform shear stress acts across thesample. The cone and plate arrangement can therefore be used to analyze nonidealmaterials.

Any of these arrangements can be used to carry out simple viscosity measurements onfluids, by measuring the variation of shear stress with shear rate. However, some of the moresophisticated ones can also be used to carry out transient and dynamic rheological tests.Typically, the rheological properties of samples are measured as a function of time ortemperature.

A number of possible sources of experimental error are associated with rheological mea-surements carried out using shear viscometers and rheometers (Sherman 1970, Hunter 1989,Pal et al. 1992). First, the gap between the cylinders or plates should be at least 20 timesgreater than the diameter of the droplets, so that the emulsion appears as a homogeneousmaterial within the device (Pal et al. 1992). On the other hand, the gap must be narrow enoughto ensure a fairly uniform shear stress across the whole of the sample. Second, a phenomenonknown as wall slip may occur within a viscometer or rheometer, which can cause seriouserrors in the measurements if not properly taken into account (Sherman 1970). It is normallyassumed that the liquid in direct contact with the surfaces of the cylinders (or plates) moveswith them at the same velocity (Hunter 1986). This assumption is usually valid for simpleliquids because the small molecules are caught within the surface irregularities on thecylinder and are therefore dragged along with it. For an emulsion, this assumption may nothold because the droplets or flocs are greater in size than the surface irregularities. Underthese circumstances, a phase separation occurs at the cylinder surface where a thin layer ofcontinuous phase acts as a lubricant and slip occurs. Wall slip effects can be taken intoaccount by roughening the surfaces of the cylinders or by using a range of different gapwidths (Hunter 1986, Pal et al. 1992). Third, the rheological properties of many samplesdepend on their previous thermal and shear history, and so this must be carefully controlledin order to obtain reproducible measurements. For example, the viscosity of many foodsdecreases substantially upon shearing due to disruption of an aggregated network of particlesor molecules, and the recovery of the initial viscosity takes a certain length of time to achieveafter the shear stress is removed. Fourth, many emulsions may be susceptible to creaming orsedimentation during the course of an experiment, which should be avoided if accuraterheological measurements are required.


8.3.3. Empirical Techniques

Many of the techniques mentioned above are unsuitable for application in the food indus-try because the instrumentation is too expensive, it requires skilled operators, or measure-ments take too long to carry out (Sherman 1970, Rao et al. 1995). For these reasons, a largenumber of empirical techniques have been developed by food scientists that provide simpleand rapid determinations of the rheological properties of a sample. Many of these empiri-cal techniques have become widely accepted for analyzing specific food types. Typicalexamples include penetrometers to measure the hardness of butters, margarines, and spreads(Sherman 1970); devices for measuring the time it takes for liquids to flow through afunnel (Liu and Masliyah 1996); or devices that measure the time it takes for a sphericalball to fall through a sample contained within a glass tube (Becher 1957). It is difficultto analyze the data from these devices using fundamental rheological concepts because itis difficult to define the stresses and strains involved. Nevertheless, these devices areextremely useful when rapid empirical information is more important than fundamentalunderstanding.

8.4. RHEOLOGICAL PROPERTIES OF EMULSIONS

Food emulsions exhibit a wide range of different rheological properties, ranging from low-viscosity liquids to fairly rigid solids. The rheological behavior of a particular food dependson the type and concentration of ingredients it contains, as well as the processing and storageconditions it has experienced. In this section, the relationship between the rheological prop-erties of emulsions and their composition and microstructure is discussed. We begin byconsidering the rheology of dilute suspensions of noninteracting rigid spheres, because thetheory describing the properties of this type of system is well established (Hiemenz 1986,Hunter 1986, Mewis and Macosko 1994, Tadros 1994). Nevertheless, many food emulsionsare concentrated and contain nonrigid, nonspherical, and/or interacting droplets (Dickinson1992). The theoretical understanding of these types of systems is less well developed,although some progress has been made, which will be reviewed.

8.4.1. Dilute Suspensions of Rigid Spherical Particles

The viscosity of a liquid increases upon the addition of rigid spherical particles because theparticles disturb the normal flow of the fluid, causing greater energy dissipation due tofriction (Hunter 1986, Mewis and Macosko 1994). Einstein derived an equation to relate theviscosity of a suspension of rigid spheres to its composition:

η η φ= +0 1 2 5( . ) (8.19)

where η0 is the viscosity of the liquid surrounding the droplets and φ is the dispersed-phasevolume fraction. This equation assumes that the liquid is Newtonian, the particles are rigidand spherical, that there are no particle–particle interactions, that there is no slip at theparticle–fluid interface, and that Brownian motion effects are unimportant. The Einsteinequation predicts that the viscosity of a dilute suspension of spherical particles increaseslinearly with particle volume fraction and is independent of particle size and shear rate. TheEinstein equation gives excellent agreement with experimental measurements for suspensionsthat conform to the above criteria, often up to particle concentrations of about 5%.

It is convenient to define a parameter known as the intrinsic viscosity of a suspension, [η](Dickinson and Stainsby 1982):


[ ]/

ηη η

φ=

−0 1(8.20)

so that

η η η φ= +0 1( [ ] ) (8.21)

For rigid spherical particles, the intrinsic viscosity tends to 2.5 as the volume fraction tendsto zero. For nonspherical particles or for particles that swell due to the adsorption of solvent,the intrinsic viscosity is larger than 2.5 (Hiemenz 1986), whereas it may be smaller for fluidparticles (Sherman 1968a,c; Dickinson and Stainsby 1982). For these systems, measurementsof [η] can provide valuable information about their shape or degree of solvation.

8.4.2. Dilute Suspensions of Fluid Spherical Particles

Food emulsions usually contain fluid, rather than solid, particles. In the presence of a flowfield, the liquid within a droplet is caused to circulate because it is dragged along by the liquid(continuous phase) that flows past the droplet (Sherman 1968a,c; Dickinson and Stainsby1982). Consequently, the difference in velocity between the materials on either side of thedroplet surface is less than for a solid particle, which means that less energy is lost due tofriction and therefore the viscosity of the suspension is lower. The greater the viscosity of thefluid within a droplet, the more it acts like a rigid sphere, and therefore the higher theviscosity of the suspension.

The viscosity of a suspension of noninteracting spherical droplets is given by (Tadros1994):

η ηη η

η ηφ= +

++

00

01

2 5. drop

drop(8.22)

where ηdrop is the viscosity of the liquid in the droplets. For droplets containing relativelyhigh-viscosity liquids (ηdrop/η0 >> 1), the intrinsic viscosity tends to 2.5, and therefore thisequation tends to that derived by Einstein (Equation 8.19). For droplets that contain relativelylow-viscosity fluids (ηdrop/η0 << 1), such as air bubbles, the intrinsic viscosity tends to unity,and so the suspension viscosity is given by η = η0(1 + φ). One would therefore expect theviscosity of oil-in-water or water-in-oil emulsions to be somewhere between these twoextremes. In practice, the droplets in most food emulsions are coated by a layer of emulsifiermolecules that forms a viscoelastic membrane. This membrane retards the transmittance ofthe tangential stress from the continuous phase into the droplet and therefore hinders the flowof the fluid within the droplet (Pal et al. 1992, Tadros 1994). For this reason, most foodemulsions contain droplets which act like rigid spheres, and so their viscosities at lowconcentrations can be described by the Einstein equation.

At sufficiently high flow rates, the hydrodynamic forces can become so large that theyovercome the interfacial forces holding the droplets together and cause the droplets to becomedeformed and eventually disrupted (Chapter 6). The shear rates required to cause dropletdisruption are usually so high that the flow profile is turbulent rather than laminar, and so itis not possible to make viscosity measurements. Nevertheless, a knowledge of the viscosityof fluids at high shear rates is important for engineers who design mixers and homogenizers.


FIGURE 8.15 Examples of oblate and prolate spheroids.

8.4.3. Dilute Suspensions of Rigid Nonspherical Particles

Many of the particles in food emulsions may have nonspherical shapes (e.g., flocculateddroplets, partially crystalline droplets, fat crystals, ice crystals, or biopolymer molecules)(Dickinson 1992). Consequently, it is important to appreciate the effects of particle shape onsuspension viscosity. The shape of many particles can be approximated as prolate spheroids(rod-like) or oblate spheroids (disk-like). A spheroid is characterized by its axis ratio rp = a/b, where a is the major axis and b is the minor axis (Figure 8.15). For a sphere, a = b; fora prolate spheroid, a > b; and for an oblate spheroid, a < b. The flow profile of a fluid arounda nonspherical particle causes a greater degree of energy dissipation than that around aspherical particle, which leads to an increase in viscosity (Hunter 1986, Hiemenz 1986,Mewis and Macosko 1994). The magnitude of this effect depends on the rotation andorientation of the spherical particle. For example, the viscosity of a rod-like particle is muchlower when it is aligned parallel to the fluid flow, rather than perpendicular, because theparallel orientation offers less resistance to flow.

The orientation of a spheroid particle in a flow field is governed by a balance between thehydrodynamic forces that act upon it and its rotational Brownian motion (Mewis and Macosko1994). The hydrodynamic forces favor the alignment of the particle along the direction of theflow field, because this reduces the energy dissipation. On the other hand, the alignment ofthe particles is opposed by their rotational Brownian motion, which favors the completerandomization of their orientations. The relative importance of the hydrodynamic and Brown-ian forces is expressed in terms of a dimensionless number, known as the Peclet number (Pe).For simple shear flow (Mewis and Macosko 1994):

Pe =«γ

DR(8.23)

where «γ is the shear rate and DR is the rotational Brownian diffusion coefficient, whichdepends on particle shape:

DkT

rR =8 3πη

for rigid spheres (8.24)

DkT

bR = 332 3πη

for circular disks (8.25)


FIGURE 8.16 At low shear rates, the particles rotate freely in all directions, but as the shear rateincreases, they become more and more aligned with the flow field. This causes a reduction in theviscosity with increasing shear rate (i.e., pseudoplasticity).

DkT

rrR p= −3

82 0 5

3πη(ln . ) for long thin rods (8.26)

When the Peclet number is much less than unity (Pe << 1), the rotational Brownian motiondominates, and the particles tend to rotate freely in the liquid. This type of behavior isobserved when the particles are small, the shear rate is low, and/or the viscosity of thesurrounding fluid is low. When the Peclet number is much greater than unity (Pe >> 1), thehydrodynamic forces dominate, and the particles become aligned with the flow field (Figure8.16). This type of behavior is observed when the particles are large, the shear rate is high,and/or the viscosity of the surrounding liquid is high.

The viscosity of a suspension of nonspherical particles therefore depends on the shearrate. At low shear rates (i.e., Pe << 1), the viscosity has a constant high value. As the shearrate is increased, the hydrodynamic forces become more important, and so the particlesbecome oriented with the flow field, which causes a reduction in the viscosity. At highshear rates (i.e., Pe >> 1), the hydrodynamic forces dominate and the particles remainaligned with the shear field, and therefore the viscosity has a constant low value (Figure8.16). Thus suspensions of nonspherical particles exhibit shear thinning behavior. The shearrate at which the viscosity starts to decrease depends on the size and shape of the particles,as well as the viscosity of the surrounding liquid. Mathematical formulae similar to Equation8.3 have been developed to calculate the influence of shear rate on the viscosity of suspen-sions of nonspherical particles, but these usually have to be solved numerically. Neverthe-less, explicit expressions are available for systems that contain very small or very largeparticles.

8.4.4. Dilute Suspensions of Flocculated Particles

When the attractive forces between the droplets dominate the repulsive forces, and aresufficiently greater than the thermal energy of the system, then droplets can aggregate intoa primary or secondary minimum (Chapter 3). The rheological properties of many food


emulsions are dominated by the fact that the droplets are flocculated, and so it is importantto understand the factors which determine the rheological characteristics of these systems. Itis often convenient to categorize systems as being either strongly flocculated (wattractive > 20kT) or weakly flocculated (1 kT < wattractive < 20 kT), depending on the strength of theattraction between the droplets (Liu and Masliyah 1996).

A dilute suspension of flocculated droplets consists of flocs which are so far apart that theydo not interact with each other through colloidal or hydrodynamic forces. This type ofsuspension has a higher viscosity than a suspension that contains the same concentration ofisolated particles because the particles in the flocs trap some of the continuous phase andtherefore have a higher effective volume fraction than the actual volume fraction (Liu andMasliyah 1996). In addition, the flocs may rotate in solution because of their rotationalBrownian motion, sweeping out an additional amount of the continuous phase and thusincreasing their effective volume fraction even more.

Suspensions of flocculated particles tend to exhibit pronounced shear thinning behavior(Figure 8.17). At low shear rates, the hydrodynamic forces are not large enough to disrupt thebonds holding the particles together, and so the flocs act like particles with a fixed size andshape, resulting in a constant viscosity. As the shear rate is increased, the hydrodynamicforces become large enough to cause flocs to become deformed and eventually disrupted. Thedeformation of the flocs results in their becoming elongated and aligned with the shear field,which results in a reduction in the viscosity. The disruption of the flocs decreases theireffective volume fraction and therefore also contributes to a decrease in the suspensionviscosity. The viscosity reaches a constant value at high shear rates, either because all of theflocs are completely disrupted so that only individual droplets remain or because the numberof flocculated droplets remains constant since the rate of floc formation is equal to that of flocdisruption (Campanella et al. 1995).

Depending on the nature of the interdroplet pair potential (Chapter 3), it is also possibleto observe shear thickening due to particle flocculation under the influence of the shear field(de Vries 1963). Some emulsions contain droplets which are not flocculated under quiescentconditions because there is a sufficiently high energy barrier to prevent the droplets fromfalling into a primary minimum. However, when a shear stress is applied to the emulsions,the frequency of collisions and the impact force between the droplets increase, which can

FIGURE 8.17 An emulsion that contains flocculated droplets exhibits shear thinning behavior be-cause the flocs are deformed and disrupted in the shear field.


cause the droplets to gain sufficient energy to “jump” over the energy barrier and becomeflocculated, therefore leading to shear thickening.

Quite complicated behavior can therefore be observed in some emulsions (Pal et al. 1992,Liu and Masliyah 1996). For example, an emulsion that contains droplets which are weaklyflocculated in a secondary minimum exhibits shear thinning at fairly low shear rates, butshows shear thickening when the shear rate exceeds some critical level where the dropletshave sufficient energy to “jump” over the energy barrier and fall into the primary minimum.The value of this critical shear rate increases as the height of the energy barrier increases. Aknowledge of the interdroplet pair potential is therefore extremely useful for understandingand predicting the rheological behavior of food emulsions.

The size, shape, and structure of flocs largely determine the rheological behavior of dilutesuspensions of flocculated particles (Dickinson and Stainsby 1982, Liu and Masliyah 1996).Flocs formed by the aggregation of emulsion droplets often have structures that are fractal(Chapter 7). The effective volume fraction (φeff) of a fractal floc is related to the size of thefloc and the fractal dimension by the following expression (Bremer 1992):

φ φeff =

−Rr

D3

(8.27)

where r is the droplet radius and R is the floc radius. The viscosity of a dilute emulsioncontaining fractal flocs can therefore be established by substituting this expression into theEinstein equation:*

η η η φ η η φ= + = + −0 0

31 1( [ ] ) [ [ ] ( / ) ]eff R r D (8.28)

If it is assumed that the flocs are approximately spherical, then [η] = 2.5. This equation offersa useful insight into the relationship between the rheology and microstructure of flocculatedemulsions.

Flocs with fairly open structures (i.e., lower D) have higher viscosities than those withcompact structures because they have higher effective volume fractions (Equation 8.28). Asmentioned earlier, the viscosity decreases with increasing shear rate partly because of disrup-tion of the flocs (i.e., a decrease in R). The shear stress at which the viscosity decreasesdepends on the magnitude of the forces holding the droplets together within a floc. Thegreater the strength of the forces, the larger the shear rate required to deform and disrupt theflocs. Thus the dependence of the viscosity of an emulsion on shear stress can be used toprovide valuable information about the strength of the bonds holding the droplets together(Sherman 1970, Dickinson and Stainsby 1982, Hunter 1989).

8.4.5. Concentrated Suspensions of Nonflocculated Particlesin the Absence of Colloidal Interactions

When the concentration of particles in a suspension exceeds a few percent, the particles beginto interact with each other through a combination of hydrodynamic and colloidal interactions,and this alters the viscosity of the system (Hunter 1986, Mewis and Macosko 1994, Tadros1994). In this section, we examine the viscosity of concentrated suspensions in the absence

* Alternatively, the expression for the effective volume fraction can be placed in the Dougherty–Krieger equationdeveloped for more concentrated emulsions (Barnes 1994).


of long-range colloidal interactions between the particles (i.e., it is assumed that the particlesact like hard spheres). The more complicated situation of suspensions in which long-rangecolloidal interactions are important is treated in the following section. Hydrodynamic inter-actions are the result of the relative motion of neighboring particles and are important in alltypes of nondilute suspensions.

At low concentrations, hydrodynamic interactions are mainly between pairs of particles,but as the concentration increases, three or more particles may be involved. As the concentra-tion increases, the measured viscosity becomes larger than that predicted by the Einsteinequation because these additional hydrodynamic interactions lead to a greater degree ofenergy dissipation. The Einstein equation can be extended to account for the effects of theseinteractions by including additional volume-fraction terms (Pal et al. 1992):

η η φ φ φ= + + + +02 31( )a b c L (8.29)

The value of the constants a, b, c, etc. can be determined either experimentally or theoreti-cally. For a suspension of rigid spherical particles, the value of a is 2.5, so that Equation 8.29tends to the Einstein equation at low-volume fractions. A rigorous theoretical treatment of theinteractions between pairs of droplets has established that b = 6.2 for rigid spherical particles(Hunter 1986). It is extremely difficult to theoretically calculate the value of higher orderterms because of the complexity of the mathematical treatment of interactions among threeor more particles. In addition, each successive constant only extends the applicability of theequation to a slightly higher volume fraction. For this reason, it has proven to be moreconvenient to adopt a semiempirical approach to the development of equations which de-scribe the viscosity of concentrated suspensions. One of the most widely used equations wasderived by Dougherty and Krieger and is applicable across the whole volume-fraction range(Hunter 1986, Mewis and Macosko 1994):

ηη

φφ

η φ

0

1= −

−

c

c[ ]

(8.30)

where φc is an adjustable parameter which is related to the volume fraction at which thespheres become closely packed and [η] is the intrinsic viscosity. Typically, the value of φc

is between about 0.6 and 0.7 for spheres which do not interact via long-range colloidalinteractions, but it may be considerably lower for suspensions in which there are strong long-range attractive or repulsive interactions between the droplets (see following sections). Theintrinsic viscosity is 2.5 for spherical particles, but may be much larger for nonspherical,swollen, or aggregated particles (Hiemenz 1986).

The viscosity of concentrated suspensions often exhibits shear thinning behavior due toBrownian motion effects (Pal et al. 1992, Mewis and Macosko 1994, Liu and Masliyah 1996).It has already been mentioned that shear thinning occurs when the shear stress is large enoughto overcome the rotational Brownian motion of nonspherical particles. Shear thinning canalso occur because of the translational Brownian motion of particles. At low shear stresses,the particles have a three-dimensional isotropic and random distribution because of theirBrownian motion (Hunter 1993). As the shear stress increases, the particles become moreordered along the flow lines to form “strings” or “layers” of particles that offer less resistanceto the fluid flow and therefore cause a decrease in the suspension viscosity.

The decrease in viscosity with increasing shear stress can be described by the followingequation:


η ηη η

τ τ= +

−+∞

∞0

1 ( / )i(8.31)

where τ i is a critical shear stress that is related to the size of the droplets (τ i = kT/βr3),and β is a dimensionless constant with a value of about 0.431 (Hunter 1989). The valueof τ i is a characteristic of a particular system which describes the relative importance ofthe translational Brownian motion and hydrodynamic shear forces. When τ << τ i , Brown-ian motion dominates and the particles have a random distribution, but when τ >> τ i, theshear forces dominate and the particles become organized into “strings” or “layers” alongthe lines of the shear field, which causes less energy dissipation. This equation indicates thatthe viscosity decreases from a constant value at low shear stresses (η0) to anotherconstant value at high shear stresses (η∞). The viscosity can decrease by as much as 30%from its low shear rate value, with the actual amount depending on the dispersed-phasevolume fraction (Hunter 1989). The shear rate at which the viscosity starts to decreasefrom its η0 value is highly dependent on the particle size. For large particles, τi is often solow that it is not possible to observe any shear thinning behavior, but for smaller particles,shear thinning behavior may be observed at the shear rates typically used in a rheologicalexperiment.

The Dougherty–Krieger equation can still be used to describe the dependence of thesuspension viscosity on dispersed-phase volume fraction, but the value of φc used in theequation is shear rate dependent. This is because droplets can pack more efficiently at highershear rates, and therefore φc increases with shear rate (Hunter 1989).

8.4.6. Suspensions of Nonflocculated Particles withRepulsive Interactions

Most food emulsions contain droplets which have various types of colloidal interactionacting between them (e.g., van der Waals, electrostatic, steric, hydrophobic, depletion, etc.)(Chapter 3). The precise nature of these interactions has a dramatic influence on the rheol-ogy of particulate suspensions. For example, two emulsions with the same droplet concen-tration could have rheological properties ranging from a low-viscosity Newtonian liquid toa highly viscoelastic material, depending on the nature of the colloidal interactions. In thissection, suspensions that contain isolated spherical particles with repulsive interactions areconsidered. In the following section, the rheological properties of flocculated emulsions areconsidered.

The major types of droplet repulsion in most food emulsions are due to electrostatic andsteric interactions. These repulsive interactions prevent the droplets from coming into closecontact when they collide with each other and therefore increase the effective volume fractionof the droplets (Tadros 1994, Mewis and Macosko 1994):

φ φ δeff = +

1

3

r(8.32)

where δ is equal to half the distance of closest separation between the two droplets.For steric stabilization, δ is approximately equal to the thickness of the adsorbed layer. For

electrostatically stabilized systems, it is related to the Debye length (κ –1) and can be describedby the following equation at low shear stresses (Mewis and Macosko 1994):


δ κ αα α

=

−1 lnln[ /(ln )]

(8.33)

where α = 4π ε0εRΨ0r 2κ exp(2rκ)/kT, ε0 is the dielectric permittivity of a vacuum, εR is therelative dielectric permittivity of the continuous phase, Ψ0 is the electrical potential at thedroplet surface, r is the radius, κ–1 is the Debye length, k is Boltzmann’s constant, and T isthe absolute temperature. For electrically charged oil droplets, the distance of closest contacttherefore decreases as the surface charge decreases or as the ionic strength of the aqueousphase increases.

It is convenient to categorize droplets with repulsive interactions as being either “hard”particles or “soft” particles (Liu and Masliyah 1996). A hard particle is incompressible, andso its effective size is independent of shear rate or droplet concentration. On the otherhand, a soft particle is compressible, and so its effective size may be reduced at high shearrates or droplet concentrations. Sterically stabilized droplets with dense interfacial layersare usually considered to act like hard particles because the layer is relatively incompress-ible, whereas electrostatically stabilized droplets or sterically stabilized droplets with openinterfacial layers are usually considered to act like soft particles because the layer iscompressible.

The viscosity of an emulsion that contains droplets with repulsive interactions can berelated to the dispersed-phase volume fraction by replacing the value of φ in the Dougherty–Krieger equation (Equation 8.33) with the effective volume fraction (φeff). A plot of viscosityversus φeff then falls on the same curve as for an emulsion that contains droplets with no long-range colloidal interactions. In emulsions that contain “soft” particles, it is also necessary toreplace the value of φc with φc,eff , to take into account that the particles may be compressedat higher volume fractions and can therefore pack more efficiently. As a consequence, theviscosity of an emulsion that contains soft particles is lower than one that contains hardparticles at the same effective volume fraction.

The influence of repulsive interactions on the rheology of emulsions depends on themagnitude of δ relative to the size of the particles. For relatively large particles (i.e., δ << r),this effect is negligible, but for small droplets or droplets with thick layers around them (i.e.,δ ≈ r), this effect can significantly increase the viscosity of a suspension (Tadros 1994).

The rheological properties of electrostatically stabilized systems that contain small emul-sion droplets are particularly sensitive to pH and salt concentration. The viscosity woulddecrease initially with increasing salt concentration because screening of the charges de-creases δ. Above a certain salt concentration, the interactions between the droplets wouldbecome attractive, rather than repulsive, and therefore flocculation will occur, causing anincrease in emulsion viscosity with salt. The rheological properties of electrostatically stabi-lized emulsions are therefore particularly sensitive to the pH, salt concentration, and type ofions present.

8.4.7. Concentrated Suspensions of Flocculated Particles

In concentrated emulsions, the flocs are close enough together to interact with each other,through hydrodynamic interactions, colloidal interactions, or entanglement. The viscosity offlocculated emulsions can be described by the Dougherty–Krieger equation by assuming thatthe flocs are “soft” particles with the actual droplet volume fraction replaced by the effectivedroplet volume fraction given by Equation 8.27. At a given actual droplet volume fraction,the emulsion viscosity increases as the size of the flocs increases or the packing of the flocsbecomes more open (lower D).


At sufficiently high droplet concentrations, flocculation may lead to the formation of athree-dimensional network of aggregated droplets (Sherman 1970, Goodwin and Ottewill1991, Pal 1996). The more open the structure of the droplets within the flocs, the lower thevalue of the actual droplet volume fraction where the network is formed. Network formationcauses a suspension of particles to exhibit plastic and/or viscoelastic characteristics (Pal1996). The network of aggregated droplets acts like a solid at low shear stresses because theapplied forces are not sufficient to overcome the forces holding the droplets together. Oncea critical shear stress is exceeded, the bonds between the droplets are disrupted and so thedroplets can flow past one another. If some of the bonds are capable of reforming during theshearing process, then the emulsion will exhibit viscoelastic behavior (Sherman 1968a). Athigher shear stresses, the rate of bond disruption greatly exceeds that of bond formation andthe emulsion acts like a liquid. Consequently, a suspension that contains a three-dimensionalnetwork of aggregated particles often has a yield stress, below which it acts like an elasticsolid and above which it acts like a liquid. Above the yield stress, the suspension oftenexhibits strong shear thinning behavior as more and more flocs are deformed and disrupted.The magnitude of the yield stress depends on the strength of the attractive forces holding theparticles together: the greater the attractive forces, the greater the yield stress (Pal 1996). Therheology of the system is also sensitive to the structural organization of the droplets (e.g.,whether they are loosely or densely packed and the number of bonds per droplet) (Bremer1992, Pal 1996).

8.4.8. Emulsions with Semisolid Continuous Phases

A number of food emulsions consist of droplets dispersed in a continuous phase which iseither partly crystalline or gelled (Sherman 1970, Dickinson and Stainsby 1982, Dickinson1992, Moran 1994). Butter and margarine consist of water droplets suspended in a liquidoil phase which contains a three-dimensional network of aggregated fat crystals. Manymeat products, desserts, and sauces consist of oil droplets suspended in an aqueous phaseof aggregated biopolymer molecules. The rheological properties of these systems are usu-ally dominated by the properties of the continuous phase, and therefore their propertiescan be described using theories developed for networks of aggregated fat crystals or biopoly-mer molecules (Clark 1987). Nevertheless, in some systems the presence of the emulsiondroplets does play a significant role in determining the overall rheological behavior (seebelow).

It is important that spreadable products, such as butter, margarine, and low-fat spreads,retain their shape when they are removed from the refrigerator, but spread easily when aknife is applied (Sherman 1970, Moran 1994). These products must therefore be designedso that they exhibit plastic properties (i.e., they have a yield stress below which they areelastic and above which they are viscous). The plastic behavior of this type of product isusually attributed to the presence of a three-dimensional network of aggregated fat crys-tals. Low shear stresses are not sufficiently large to disrupt the bonds which hold theaggregated crystals together, and so the product exhibits solid-like behavior. Above theyield stress, the applied shear stress is sufficiently large to cause the bonds to be disrupted,so that the fat crystals flow over each other and the product exhibits viscous-like behavior.After the stress is removed, the bonds between the fat crystals reform over time, andtherefore the product regains its elastic behavior. The creation of a product with the desiredrheological characteristics involves careful selection and blending of various food oils, aswell as control of the cooling and shearing conditions used during the manufacture of theproduct. To the author’s knowledge, little systematic research has been carried out toestablish the influence of the characteristics of the water droplets on the rheological prop-


erties of these products (e.g., particle size distribution, emulsifier type, and dispersed-phasevolume fraction).

A great deal of research has recently been carried out to determine the influence of oildroplets on the rheology of filled gels (Jost et al. 1986; Aguilera and Kessler 1989; Xiong etal. 1991; Xiong and Kinsella 1991; Yost and Kinsella 1993; McClements et al. 1993c;Dickinson and Hong 1995a,b, 1996; Dickinson and Yamamoto 1996). These filled gels arecreated by heating oil-in-water emulsions which contain a significant amount of whey proteinin the continuous phase above a temperature where the proteins unfold and form a three-dimensional network of aggregated molecules. The oil droplets can act as either structurepromoters or structure breakers, depending on the nature of their interaction with the gelnetwork. When the droplets are stabilized by dairy proteins, the attractive interactions be-tween the adsorbed proteins and those in the network reinforce the network and increase thegel strength. Conversely, when the droplets are stabilized by small-molecule surfactants(which do not interact strongly with the protein network), the presence of the droplets tendsto weaken the network and decrease the gel strength. Quite complex rheological behavior canbe observed in emulsions that contain mixtures of proteins and small-molecule surfactants(Dickinson and Hong 1995a,b; Dickinson and Yamamoto 1996). The shear modulus of filledgels that contain protein-stabilized oil droplets increases dramatically when a small amountof surfactant is added to the system, but decreases at higher surfactant values. The incorpo-ration of surfactants into filled gels may therefore prove to be an effective means of control-ling their rheological properties.

The influence of the emulsion droplets also depends on their size relative to the pore sizeof the gel network (McClements et al. 1993c, Yost and Kinsella 1993). If the droplets arelarger than the pore size, they tend to disrupt the network and decrease the gel strength, butif they are smaller than the pore size, they are easily accommodated into the network withoutdisrupting it.

8.5. MAJOR FACTORS THAT DETERMINE EMULSION RHEOLOGY

8.5.1. Dispersed-Phase Volume Fraction

The viscosity of an emulsion increases with dispersed-phase volume fraction. At low dropletconcentrations, this increase is linearly dependent on volume fraction (Equation 8.19), but itbecomes steeper at higher concentrations (Equation 8.30). Above a critical dispersed-phasevolume fraction (φc), the droplets are packed so closely together that they cannot easily flowpast each other, and so the emulsion has gel-like properties. The precise nature of thedependence of the viscosity on volume fraction is mainly determined by the nature of thecolloidal interactions between the droplets.

8.5.2. Rheology of Component Phases

The viscosity of an emulsion is directly proportional to the viscosity of the continuous phase(Equations 8.19 and 8.30), and so any alteration in the rheological properties of the continu-ous phase has a corresponding influence on the rheology of the whole emulsion. For thisreason, the presence of a thickening agent in the aqueous phase of an oil-in-water emulsion(Pettitt et al. 1995, Pal 1996) or the presence of a fat crystal network in the oil phase of awater-in-oil emulsion (Moran 1994) largely determines the overall rheological properties ofthe system.

The rheology of the dispersed phase has only a minor influence on the rheology of foodemulsions because the droplets are covered with a fairly viscoelastic membrane, which meansthey have properties similar to rigid spheres (Tadros 1994, Walstra 1996a).


8.5.3. Droplet Size

The influence of the droplet size and the droplet size distribution on the rheology of anemulsion depends on the dispersed-phase volume fraction and the nature of the colloidalinteractions. The viscosity of dilute emulsions is independent of the droplet size when thereare no long-range attractive or repulsive colloidal interactions between the droplets (Pal et al.1992, Pal 1996). When there is a relatively long-range repulsion between the droplets (i.e.,δ ∼ r ), their effective volume fraction is much greater than their actual volume fraction, andso there is a large increase in the viscosity of the emulsion (Tadros 1994, Pal 1996). Thedroplet size also influences the degree of droplet flocculation in an emulsion (Chapters 3 and7), which has an impact on the emulsion rheology. For example, the greater the extent ofdroplet flocculation, or the more open the structure of the flocs formed, the larger theemulsion viscosity.

The mean droplet size and degree of polydispersity have a particularly significant influenceon the rheology of concentrated emulsions (Liu and Masliyah 1996, Pal 1996). In emulsionsthat contain nonflocculated droplets, the maximum packing factor (φc) depends on the poly-dispersity. Droplets are able to pack more efficiently when they are polydisperse, and there-fore the viscosity of a concentrated polydisperse emulsion is less than that of a monodisperseemulsion with the same droplet volume fraction. Emulsions that contain flocculated dropletsare able to form a three-dimensional gel network at lower volume fractions when the dropletsize decreases.

The droplet size also alters the rheology of emulsions due to its influence on the relativeimportance of rotational and translation Brownian motion effects compared to the shear stress(Mewis and Macosko 1994).

8.5.4. Colloidal Interactions

The nature of the colloidal interactions between the droplets in an emulsion is one of the mostimportant factors determining its rheological behavior. When the interactions are long rangeand repulsive, the effective volume fraction of the dispersed phase may be significantlygreater than its actual volume fraction, φeff = φ(1 + δ/r)3, and so the emulsion viscosityincreases (Section 8.4.6). When the interactions between the droplets are sufficiently attrac-tive, the effective volume fraction of the dispersed phase is increased due to droplet floccu-lation, which results in an increase in emulsion viscosity (Section 8.4.4). The rheologicalproperties of an emulsion therefore depend on the relative magnitude of the attractive (mainlyvan der Waals, hydrophobic, and depletion interactions) and repulsive (mainly electrostatic,steric, and thermal fluctuation interactions) interactions between the droplets (Chapter 3). Afood scientist can therefore control the rheological properties of food products by manipulat-ing the colloidal interactions between the droplets. Increases in the viscosity of oil-in-wateremulsions due to droplet flocculation have been induced by adding biopolymers to increasethe depletion attraction (Dickinson and Golding 1997a,b), by adding biopolymers to causebridging flocculation (Dickinson and Golding 1997a), by altering the pH or ionic strength toreduce electrostatic repulsion (Hunt and Dalgleish 1994, Demetriades et al. 1997a), and byheating protein-stabilized emulsions to increase hydrophobic attraction (Demetriades et al.1997b).

8.5.5. Particle Charge

The charge on an emulsion droplet can influence the rheological properties of an emulsionin a number of ways. First, the charge determines whether the droplets are aggregated orunaggregated and the distance of closest approach (Section 8.5.4). Second, the dropletcharge influences the rheology due to the primary electroviscous effect (Pal 1996). As a


FIGURE 8.18 The primary electroviscous effect increases the viscosity of an emulsion.

charged droplet moves through a fluid, the cloud of counterions surrounding it becomesdistorted (Figure 8.18). This causes an attraction between the charge on the droplet and thatassociated with the cloud of counterions that lags slightly behind it. This attraction opposesthe movement of the droplets and therefore increases the emulsion viscosity because moreenergy is needed to cause the droplets to move at the same rate as uncharged droplets.


9 Appearance and Flavor

9.1. INTRODUCTION

The modern consumer is faced with a huge variety of different types of food products fromwhich to choose, and within each category there are a number of different brand names. Theinitial choice of a particular product is governed by many factors, including its cost, quality,packaging, ease of preparation, and nutritional value. Once a consumer has made a decisionto purchase a certain brand name, the manufacturer wants to ensure that he or she issatisfied with the product and will purchase it again. It is therefore important to make certainthat the product has desirable and reproducible quality attributes (i.e., appearance, taste,texture, and shelf life). In previous chapters, we considered emulsion rheology, which isrelated to texture (Chapter 8), and emulsion stability, which is related to shelf life (Chapter7). In this chapter, focus is on the factors which influence the flavor and appearance of foodemulsions.

9.2. EMULSION FLAVOR

The term “flavor” refers to those volatile components in foods which are sensed by receptorsin the nose (aroma) and those nonvolatile components which are sensed by receptors in thetongue and the inside of the mouth (taste) (Thomson 1986). In addition, certain componentsin foods may also contribute to flavor because of their influence on the perceived texture(mouthfeel) (Kokini 1987). The flavor of a food is therefore a combination of aroma, taste,and mouthfeel, with the former usually the most important (Taylor and Linforth 1996).Flavor perception is an extremely complicated process which depends on a combination ofphysicochemical, biological, and psychological phenomena (Thomson 1986, Bell 1996).Before a food is placed in the mouth, its flavor is perceived principally through those volatilecomponents which are inhaled directly into the nasal cavity. After the food is placed in themouth, the flavor is determined by nonvolatile molecules which leave the food and aresensed by receptors on the tongue and the inside of the mouth, as well as by those volatilemolecules which are drawn into the nasal cavity through the pharynx at the back of themouth (Thomson 1986). The interactions between flavor molecules and human receptorswhich lead to the perceived flavor of a food are extremely complicated and are still fairlypoorly understood (Thomson 1986). In addition, expectations and eating habits vary fromindividual to individual, so that the same food may be perceived as tasting differently by twoseparate individuals or by a single individual at different times. This section focuses on thephysicochemical aspects of flavor partitioning and release in foods, because these are themost relevant topics to emulsion science. Although the physiological and psychologicalaspects of flavor are extremely important, they are beyond the scope of this book.

It is widely recognized that the perceived flavor of a food is not simply determined by thetype and concentration of the flavor molecules present (Taylor and Linforth 1996), but by anumber of other physicochemical factors as well, including:


1. The environment (matrix) in which the flavor molecules are located (e.g., the lipid,aqueous, or interfacial regions)

2. The physical state of the environment (e.g., gas, liquid or solid)3. The structural organization of the components (e.g., emulsion versus nonemulsion)4. The chemical state of the flavor molecules (e.g., degree of ionization or self-

association)5. The physical and chemical interaction of flavors with other molecules (e.g., pro-

teins, carbohydrates, surfactants, or minerals)6. The rate at which flavor molecules move from one environment to another

Given the large number of factors which contribute to food flavor, it is extremely difficult toaccurately predict the flavor of a final product from a knowledge of its physicochemicalcharacteristics. For this reason, the formulation of food flavors is often the result of art andcraft, rather than the application of fundamental scientific principles. Even so, a more rigor-ous scientific approach to this topic would have great benefits for the food industry becauseit would enable manufacturers to design foods in a more systematic and cost-effectivemanner. In this section, some of the more fundamental aspects of the science of emulsionflavor are reviewed.

9.2.1. Flavor Partitioning

The perception of a flavor depends on the precise location of the flavor molecules withinan emulsion. Most flavors are perceived more intensely when they are present in theaqueous phase rather than the oil phase (McNulty 1987, Kinsella 1989). The aroma isdetermined by the presence of volatile molecules in the vapor phase above an emulsion(Overbosch et al. 1991, Taylor and Linforth 1996). Certain flavor molecules may associatewith the interfacial region, which alters their concentration in the vapor and aqueous phases(Wedzicha 1988). It is therefore important to establish the factors which determine thepartitioning of flavor molecules within an emulsion. An emulsion system can be conve-niently divided into four phases between which the flavor molecules distribute themselves:the interior of the droplets, the continuous phase, the oil–water interfacial region, and thevapor phase above the emulsion.* The relative concentration of the flavor molecules ineach of these regions depends on their molecular structure and the properties of each of thephases (Baker 1987, Bakker 1995). In this section, we start by examining flavor partitioningin some simple model systems and then move on to some more complicated and realisticmodel systems. It should be stressed that most of the physical principles of flavor partition-ing in emulsions are also applicable to the partitioning of other types of food ingredients,such as antioxidants, colors, preservatives or vitamins (Wedzicha 1988, Coupland andMcClements 1996, Huang et al. 1997).

9.2.1.1. Partitioning in Homogenous Liquids

The simplest situation to consider is the partitioning of a flavor between a homogeneousliquid and the vapor above it (Figure 9.1). At thermodynamic equilibrium, the flavor distrib-utes itself between the liquid and vapor according to the equilibrium partition coefficient(Wedzicha 1988):

* Flavors could also be present at the air–emulsion interface, but the interfacial area of this region is usually so smallthat the amount of flavor involved is negligible.


FIGURE 9.1 Flavor molecules distribute themselves between a bulk liquid and its vapor phaseaccording to their equilibrium partition coefficient (KGL = cG/cL).

Ka

aGLG

L= (9.1)

where aG and aL are the activity coefficients of the flavor in the gas and liquid phases,respectively. The concentration of flavors in foods is usually very low, and so the activitycoefficients can be replaced by concentrations, since interactions between flavor moleculesare insignificant (Wedzicha 1988, Overbosch et al. 1991):

Kc

cGLG

L= (9.2)

where cG and cL are the concentrations of the flavor in the gas and liquid phases, respectively.It is often more convenient to represent the partitioning of a flavor as the mass fraction of thetotal amount of flavor in the system which is in the vapor phase:

ΦmL G GLV V K

=+

11 / ( ) (9.3)

where VG and VL are the volumes of the gas and liquid phases, respectively.The magnitude of the partition coefficient depends on the relative strength of the interac-

tions between a flavor molecule and its surroundings in the gas and liquid phases (Tinoco etal. 1985, Israelachvili 1992):

KG

RTGLGL= −

exp

∆(9.4)

where ∆GGL is the difference in free energy per mole of the solute in the gas and liquid phases.The free energy term depends on the change in molecular interactions and configurational

entropy which occurs when a flavor molecule moves from the vapor phase into the liquidphase. The configurational entropy favors the random distribution of the flavor moleculesthroughout the whole volume of the system, rather than their confinement to just the liquidphase, and therefore it favors volatilization. The change in energy associated with the molecu-lar interactions is determined by the formation of solvent–flavor bonds (≈zwSF) and thedisruption of solvent–solvent bonds (≈ 1

2 zwSS), which occurs when a flavor molecule moves


TABLE 9.1Equilibrium Partition Coefficients of Some CommonFlavors in Oil and Water at 25°C

Air–water Air–oilCompound (KGW) (KGO)

AldehydesAcetaldehyde 2.7 × 10–3

Propanal 3.0 × 10–3

Butanal 4.7 × 10–3 2.3 × 10–3

Pentanal 6.0 × 10–3 1.0 × 10–3

Hexanal 8.7 × 10–3 0.35 × 10–3

Heptanal 11 × 10–3 0.10 × 10–3

Octanal 21 × 10–3 0.04 × 10–3

Nonanal 30 × 10–3

KetonesButan-2-one 1.9 × 10–3 1.9 × 10–3

Pent-2-one 2.6 × 10–3

Heptan-2-one 5.9 × 10–3 1.03 × 10–3

Oct-2-one 7.7 × 10–3

Nonan-2-one 15 × 10–3

Undecan-2-one 26 × 10–3

AlcoholsMethanol 0.18 × 10–3

Ethanol 0.21 × 10–3 10 × 10–3

Propanol 0.28 × 10–3

Butanol 0.35 × 10–3 0.57 × 10–3

Pentanol 0.53 × 10–3

Hexanol 0.63 × 10–3 0.36 × 10–3

Heptanol 0.77 × 10–3

Octanol 0.99 × 10–3 0.09 × 10–3

Data compiled from Buttery et al. (1969, 1971, 1973) and Overbosch et al.(1991).

from the vapor into the solvent.* Here, z is the coordination number of the flavor molecules,and wSS and wSF are the solvent–solvent and solvent–flavor interaction energies (Chapter 2).The overall change therefore depends on the relative strength of both the solvent–solvent andsolvent–flavor interactions: ∆w ≈ zwSF – 1

2 zwSS.The molecular characteristics of flavor molecules largely determine their partition coeffi-

cient in different solvents (Table 9.1). When a nonpolar flavor molecule is dispersed in anonpolar solvent, or when a polar flavor molecule is dispersed in a polar solvent, there is adecrease in volatility with increasing molecular weight (Buttery et al. 1973). This is becausethe number of favorable attractive interactions between the flavor molecules and the solventincreases as the size of the flavor molecules increases. On the other hand, when a nonpolarflavor molecule is dispersed in a polar solvent, there is a decrease in volatility with increasingmolecular weight (Buttery et al. 1969, 1971; Bomben et al. 1973; Franzen and Kinsella 1975).Nonpolar flavors are less volatile in nonpolar solvents than in polar solvents (Buttery et al.1973, Bakker 1995), because a number of relatively strong hydrogen bonds have to be

* This simple analysis assumes that the sizes of the solvent and flavor molecules are approximately equal.


replaced by relatively weak van der Waals bonds when a nonpolar molecule is introduced intoa polar solvent, which is energetically unfavorable because of the hydrophobic effect. Inaddition, polar flavors are less volatile in polar solvents than in nonpolar solvents, becausethe hydrogen bonds which form in polar solvents are more strongly attractive than the vander Waals bonds which form in nonpolar solvents.

9.2.1.2. Influence of Flavor Ionization

A number of water-soluble flavors have chemical groups which are capable of undergoingproton association–dissociation as a result of changes in pH (e.g., –COOH → –COO– + H+

or –NH3+ → –NH2 + H+). The volatility and flavor characteristics of the different ionic forms

of a molecule are different because of changes in their molecular interactions with the solvent(Baldwin et al. 1973, Wedzicha 1988, Guyot et al. 1996). For example, the ionized form ofa flavor is less volatile in an aqueous solution than the nonionized form because of strongion–dipole interactions between it and the surrounding water molecules. It is thereforeimportant to take into account the effect of ionization on the partitioning of flavor molecules.The concentration of a specific ionic form at a certain pH can be determined using theHenderson–Hasselbach equation: pH = pKa – log(cAcid/cBase), where pKa = –log(Ka) and Ka

is the dissociation constant of the acidic group (Atkins 1994).The volatility of the ionized form of a flavor is usually much lower than that of the

nonionized form, and so the concentration of the flavor in the vapor phase is determinedprincipally by the amount of nonionized flavor (cL,N) present in the liquid phase. The partitioncoefficient is therefore given by:

Kc

c

c

cGLG

L N

G

L

= =+ − −

, ( )1 10 1pH pKa (9.5)

In practice, it is more convenient to define an effective partition coefficient, which is equalto the concentration of the flavor in the vapor phase relative to the total amount of flavor inthe liquid phase (cL = cL,I + cL,N), where cL,I is the concentration of the ionized form of theflavor:

Kc

c

KGe

LG

L

GL= =+ −( )1 10pH pKa

(9.6)

When the pH of the aqueous solution is well below the pKa value of the acid group, theflavor molecule is almost exclusively in the nonionized form (Ke

GL = KGL), and so the flavorin the vapor phase is at its most intense. As the pH is raised toward the pKa value of the acidgroup, the fraction of flavor molecules in the nonionized form decreases (Ke

GL < KGL), and sothe flavor in the vapor phase becomes less intense.

It should be stressed that the ionization of a flavor molecule may also influence thepartition coefficient because it alters its interactions with other charged molecules within theaqueous phase. For example, there may be attractive electrostatic interactions between anionized flavor molecule and an oppositely charged biopolymer, which leads to flavor bindingand therefore a reduction of its concentration in the vapor phase (Section 9.2.1.3).

9.2.1.3. Influence of Flavor Binding

Many proteins and carbohydrates are capable of binding flavor molecules and thereforealtering their distribution within an emulsion (Franzen and Kinsella 1975, Bakker 1995,


O’Neill 1996, Hansen and Booker 1996, Hau et al. 1996). Flavor binding can cause asignificant alteration in the perceived flavor of a food. This alteration is often detrimental tofood quality because it changes the characteristic flavor profile, but it can also be beneficialwhen the bound molecules are off-flavors. A flavor chemist must therefore take bindingeffects into account when formulating the flavor of a particular product.

The equilibrium partition coefficient of a flavor in the presence of a biopolymer is givenby:

Kc

cGLG

L F

=,

(9.7)

where cL,F is the concentration of free (unbound) flavor present in the liquid. The concentra-tion of free flavor depends on the nature of the binding between the flavor and biopolymer(Overbosch et al. 1991). Binding may be the result of covalent bond formation or physicalinteractions, such as electrostatic, hydrophobic, van der Waals, or hydrogen bonds (Chapter2). It may take place at specific sites on the surface of a biopolymer molecule or nonspecificallyat any location on the surface. It may be reversible or irreversible.

The extent of flavor binding to a biopolymer can be conveniently characterized by abinding coefficient:

Kc

cL B

L F*

,

,= (9.8)

where cL,B is the concentration of bound flavor in the liquid phase (cL = cL,F + cL,B). Thestronger the binding between the flavor and a biopolymer, the greater the value of K*.

It is convenient to define an effective partition coefficient, which relates the concentrationof flavor in the gas phase to the total amount of flavor in the liquid phase (Overbosch et al.1991):

Kc

c

K

KGe

LG

L

GL= =+1 * (9.9)

This equation indicates that the concentration of flavor in the vapor phase is not influencedby the biopolymer when the binding coefficient is small (K* << 1), but that there is a largereduction in the concentration in the vapor phase when the binding is strong (K* > 1).

Binding constants are frequently measured experimentally by equilibrium dialysis. Abiopolymer solution is placed inside a semipermeable dialysis bag, which is then suspendedin a solution of the flavor molecules (Figure 9.2). The large biopolymer molecules arerestricted to the inside of the dialysis bag, while the small flavor molecules can move throughit. After the system has reached equilibrium, the amount of flavor bound to the biopolymermolecules is determined by measuring the concentration of flavor in the bag above that whichwould be expected in the absence of biopolymer. An alternative technique which is alsowidely used to measure flavor binding is head space analysis (Section 9.2.3). In this tech-nique, the partition coefficient is determined by measuring the equilibrium concentration ofvolatiles in the head space above a solution. By measuring the partition coefficient at differentbiopolymer concentrations, it is possible to determine the binding constant.

Flavor molecules, such as alkanes, aldehydes, and ketones, have been shown to be capableof specifically binding to various different types of protein (e.g., casein, β-lactoglobulin,


bovine serum albumin, and soy protein) (Langourieux and Crouzet 1995, O’Neill 1996,Hansen and Booker 1996, Boudaud and Dumont 1996). These proteins have little flavorthemselves, but can cause significant changes in the flavor profile of an emulsion by bindingeither desirable or undesirable flavors. The extent of flavor binding depends on the molecularstructure of the flavor and protein molecules. Nonpolar flavors are believed to bind tononpolar patches on the surfaces of proteins through hydrophobic attraction. An increase inbinding of flavors to β-lactoglobulin has been observed in the presence of urea or on heating,because these treatments cause partial unfolding of the proteins, which increases their surfacehydrophobicity (O’Neill 1996). Flavors may also bind to polysaccharides, but in this case themolecular interactions involved are more likely to be van der Waals, electrostatic, or hydro-gen bonds (Hau et al. 1996). In some systems, there may be irreversible binding due tochemical reactions between biopolymer and flavor molecules.

9.2.1.4. Influence of Surfactant Micelles

Surfactants are normally used to physically stabilize emulsion droplets against aggregation byproviding a protective membrane around the droplet (Chapter 7). Nevertheless, there is oftenenough free surfactant present in an aqueous phase to form surfactant micelles (Section 4.5).These surfactant micelles are capable of solubilizing nonpolar molecules in their hydrophobicinterior, which increases the affinity of nonpolar flavors for the aqueous phase and thereforedecreases their partition coefficient (KGL). By a similar argument, reverse micelles in an oilphase are capable of solubilizing polar flavor molecules and therefore decreasing theirpartition coefficient (KGL).

The partitioning of flavor compounds between a surfactant solution and the gas above itcan be described by the following equation:

1K K KGL

M

GM

C

GC

= +φ φ

(9.10)

where φM is the volume fraction of the micelles, φC is the volume fraction of the continuousphase surrounding the micelles (φM + φC = 1), KGM (= cG/cM) is the gas–micelle partitioncoefficient, and KGC (= cG/cC) is the gas–continuous phase partition coefficient. It is difficultto directly measure the partitioning of a flavor between gas and micelles, and so it is moreconvenient to rewrite the above equation in the following form:

FIGURE 9.2 The principles of equilibrium dialysis. A biopolymer solution is placed in a dialysis bagwhich is then suspended in a flavor solution. The amount of flavor bound to the biopolymer moleculesis determined by analyzing the flavor concentration in the cell.


KK

KGLGC

M MC

=+ −[ ( )]1 1φ (9.11)

where KMC (= cM/cC) is the partition coefficient of the flavor molecules between the micellesand the continuous phase. Equation 9.11 indicates that the higher the affinity of the flavormolecules for the micelles (KMC >> 1), the smaller the concentration of flavor above thesolution.

9.2.1.5. Partitioning in Emulsions in the Absence ofan Interfacial Membrane

In an emulsion, which consists of two immiscible liquids, we must consider the partitioningof the additive between the emulsion droplets and the surrounding liquid, as well as into thevapor phase (Figure 9.3). We therefore have to define three different partition coefficients:

Kc

cK

c

cK

c

cDCD

CGC

G

CGD

G

D

= = = (9.12)

where K is the partition coefficient; c is the concentration; and the subscripts D, C, and G referto the dispersed, continuous, and gas phases, respectively. The partitioning of the additivebetween the oil and aqueous phases depends on the relative strength of its molecular inter-actions with its surroundings in the two different phases. Thus nonpolar molecules will tendto exist preferentially in the oil phase, while polar molecules will tend to exist preferentiallyin the aqueous phase. Thus, for oil-in-water emulsions, KDC is large for nonpolar moleculesand small for polar molecules, while the opposite is true for water-in-oil emulsions.

The partition coefficient of a flavor molecule between an emulsion and vapor phase isgiven by (Overbosch et al. 1991):

1K K KGE

D

GD

C

GC

= +φ φ

(9.13)

where φD + φC = 1. Thus the partition coefficient between an emulsion and its vapor can bepredicted from a knowledge of KGD and KGC. Experiments with flavor compounds dispersed

FIGURE 9.3 In a two-liquid system, the flavor partitions between the oil, water, and vapor phasesaccording to the partition coefficients.


FIGURE 9.4 Influence of dispersed-phase volume fraction on the concentration of a polar (KDC =0.01) and a nonpolar (KDC = 100) flavor in the vapor phase of an oil-in-water emulsion (VG = 10 cm3,VE = 100 cm3).

in oil-in-water emulsions have shown that this equation gives a good description of thebehavior of emulsions, provided the flavor does not interact with the interface or any freeemulsifier in the aqueous phase (Guyot et al. 1996).

Predictions of the mass fraction of flavor in the vapor phase of oil-in-water emulsions withthe same overall flavor concentration but different dispersed-phase volume fractions areshown in Figure 9.4. There is a decrease in the fraction of a nonpolar flavor in the vapor phaseas the oil content increases, whereas the amount of a polar flavor is relatively unaffected.Thus the nonpolar flavors in an emulsion become more odorous as the fat content is de-creased, whereas the polar flavors remain relatively unchanged. This has important conse-quences when deciding the type and concentration of flavors to use in low-fat analogs ofexisting emulsion-based food products.

One possible limitation of Equation 9.13 is that it does not take into account the dropletsize. The assumption that the partitioning of additives is independent of particle size is likelyto be valid for emulsions which contain fairly large droplets (i.e., d > 1 µm), because theinfluence of the curvature of a droplet on the solubility of the material within it is notsignificant (Hunter 1986). However, when the droplet diameter falls below a critical size,there is a large increase in the solubility of the material within it because of the increasedLaplace pressure. Thus one would expect the flavor concentration in the continuous phaseand in the vapor phase to increase as the size of the droplets in an emulsion decreased. Thesepredictions are supported by experimental evidence which indicates that the partition coef-ficient of a nonpolar additive in an emulsion is less than in a macroscopic two-phase systemwith the same overall composition (Matsubara and Texter 1986, Texter et al. 1987, Wedzicha1988).

9.2.1.6. Partitioning in Emulsions in the Presence ofan Interfacial Membrane

Even though the interfacial region constitutes only a small fraction of the total volume of anemulsion, it can have a pronounced influence on the partitioning of surface-active molecules,especially when they are present at low concentrations, which is usually the case for foodflavors. The influence of the interfacial membrane can be highlighted by a simple calculationof the amount of a surface-active additive which can associate with it (Wedzicha 1988).Assume that the additive occupies an interfacial area of 1 m2/mg, which is typical of many


surface-active components (Dickinson 1992). The interfacial area per unit volume of anemulsion is given by the following relationship: AS = 6φ/dVS, where dVS is the volume–surfacemean diameter (McClements and Dungan 1993). If we assume that the additive is present in100 cm3 of an emulsion with a dispersed-phase volume fraction of 0.1 and a mean dropletdiameter of 1 µm, then the total interfacial area of the droplets is 60 m2. It would thereforetake about 60 mg of additive to completely saturate the interface, which corresponds to aconcentration of approximately 0.1 wt%. Many flavors are used at concentrations which areconsiderably less than this value, and therefore their ability to accumulate at an interface hasa large influence on their partitioning within an emulsion.

The accumulation of a flavor at an interface reduces its concentration in the oil, water, andgaseous phases by an amount which depends on the interfacial area, the flavor concentration,and the affinity of the flavor for the interface.

Reversible Binding. When the binding between the flavor and the interface is reversible, wecan define a number of additional partition coefficients:

Kc

cK

c

cK

c

cIDI

DIC

I

CIG

I

G

= = = (9.14)

In this case, the partition coefficient between the gas and the emulsion is given by:

1K K K KGE

D

GD

C

GC

L

GI

= + +φ φ φ

(9.15)

In practice, it is difficult to directly measure the partition coefficient between the gas andinterfacial region (KGI), and so it is better to express the equation in terms of properties whichare simpler to measure (i.e., KGI = KGC/KIC). In addition, the properties of the interface areusually better expressed in terms of the interfacial area rather than the volume fraction.Equation 9.15 can therefore be expressed in the following manner:

1K K K

A K

KGE

D

GD

C

GC

S IC

GC

= + +φ φ *

(9.16)

where K*IC = ΓI /cC is the partition coefficient between the interface and the continuous phase

and ΓI is the mass of the additive per unit interfacial area. Thus the partition coefficient ofan emulsion (KGE) can be predicted from experimental measurements which are all relativelysimple to carry out (i.e., KGC, KGD, and KIC). This equation assumes that the concentration offlavor at the interface is well below the saturation level. Once the interfacial region becomessaturated with flavor, the remainder will be distributed between the bulk phases.

The influence of the interface on the volatility of a surface-active flavor molecule is shownin Figure 9.5. As the size of the droplets in the emulsion is decreased, the interfacial areaincreases, and therefore a greater amount of flavor associates with the interface, therebyreducing its volatility. Nevertheless, it should be stated that this type of behavior is only likelyto occur when there is no free emulsifier in the aqueous phase, either as individual moleculesor micelles. In practice, there is often free emulsifier in the aqueous phase and the flavormolecules may bind to it as well as to the interfacial region. In these systems, the flavorvolatility is more likely to depend on the total concentration of emulsifier in the system ratherthan on the droplet size. Another factor which must be considered is that the concentration


FIGURE 9.5 Influence of droplet size on the volatility of surface-active flavor molecules whichassociate with the interfacial membrane.

of a flavor in the continuous and gas phases may actually increase with decreasing dropletsize because of the influence of the droplet curvature on solubility (Section 9.2.1.5).

Irreversible Binding. When the binding of the flavor to the interface is irreversible, then itsconcentration in the vapor phase is only determined by the amount of free flavor in theemulsion (KGE = cG/cE,F). Under these circumstances, it is usually more convenient to use aneffective partition coefficient, which is equal to the concentration of flavor in the vapor phaserelative to the total amount of flavor in the emulsion (Ke

GE = cG/cE):

1K

c

c A K KGEe

E

E S I

D

GD

C

GC

=−

+

Γ

φ φ(9.17)

where, ΓI is now the amount of flavor that is irreversibly bound to the interface per unitsurface area. If the flavor does not interact with the interface, and the interfacial region hasa negligible volume, then this equation reduces to Equation 9.13, but if the flavor binds tothe interface, its concentration in the vapor phase is reduced.

Recent studies using oil-in-water emulsions that contain different types of flavor com-pounds have indicated that amphiphilic flavors, such as butyric acid, bind strongly to theinterface of a droplet and thus reduce the partition coefficient (KGL) (Guyot et al. 1996).Nevertheless, a great deal of systematic research is still needed to determine the factors whichinfluence the volatility of different flavor compounds in food emulsions. Special emphasisshould be placed on establishing the molecular basis of this process so that predictions aboutthe flavor profile of a food can be made from a knowledge of its composition and the typeof flavor components present. This type of information could then be used by flavor chemiststo formulate foods with specific flavor profiles.

9.2.2. Flavor Release

Flavor release is the process whereby flavor molecules move out of a food and into thesurrounding saliva or vapor phase during mastication (McNulty 1987, Overbosch et al. 1991).The release of the flavors from a food material occurs under extremely complex and dynamicconditions (Land 1996). A food usually spends a relatively short period (typically 1 to 30 s)in the mouth before being swallowed. During this period, it is diluted with saliva, experiences


temperature changes, and is subjected to a variety of mechanical forces. Mastication maytherefore cause dramatic changes in the structural characteristics of a food.

During mastication, nonvolatile flavor molecules must move from within the food, throughthe saliva, to the taste receptors on the tongue and the inside of the mouth, whereas volatileflavor molecules must move from the food, through the saliva, and into the gas phase, wherethey are carried to the aroma receptors in the nasal cavity (Thomson 1986). The two majorfactors which determine the rate at which these processes occur are the equilibrium partitioncoefficient (because this determines the magnitude of the flavor concentration gradients at thevarious boundaries) and the mass transfer coefficient (because this determines the speed atwhich the molecules move from one location to another). In this section, we examine someof the simple models which have been developed to describe the complex processes whichoccur during the release of both nonvolatile and volatile flavor components from foods.

9.2.2.1. Release of Nonvolatile Compounds (Taste)

Ideally, we would like to know the maximum amount of flavor which can be released froma food and the time taken for this release to occur.

Maximum Amount Released. A relatively simple model, based on the equilibrium partitioncoefficient of the flavor between oil and water, has been used to describe the maximumamount of flavor which can be released by an oil-in-water emulsion when it is placed in themouth (McNulty 1987). The model assumes that the food is initially at equilibrium, so thatthe distribution of the flavor between the droplets and continuous phase is given by theequilibrium partition coefficient (Kow). When the food is placed in the mouth, it is diluted bysaliva (Figure 9.6). Immediately after dilution, the concentration of flavor in the aqueousphase is reduced, and so there is a thermodynamic driving force which favors the release offlavor from the droplets until the equilibrium flavor distribution is restored.

The potential extent of the flavor release can be characterized by the ratio of the flavor inthe aqueous phase once equilibrium has been reestablished to that immediately after dilution:

Ec

c

K DF

K DFFWe

Wd

ow

ow

= =− + −− + −

[ ( ) ]( )

[ ( ) ]( )

φ φφ φ

1 1

1 1 (9.18)

where DF is the dilution factor of the emulsion (= Vf /Vi), Vi and Vf are the emulsion volumebefore and after dilution, φ is the dispersed-phase volume fraction of the initial emulsion, cWd

is the concentration of flavor in the aqueous phase immediately after dilution, and cWe is theconcentration in the aqueous phase once equilibrium has been reestablished. The higher the

FIGURE 9.6 The flavor in a food is initially distributed according to the partition coefficients. Whenit is diluted with saliva, the equilibrium is upset, and flavor is released from the droplets.


value of EF, the greater the potential for flavor release. Despite its simplicity, this model canbe used to make some valuable predictions about the factors which determine the flavorrelease from foods (McNulty 1987) (e.g., the extent of flavor release on dilution increases aseither φ or Kow increases). The major limitation of this model is that it provides no informationabout the rate at which the flavor is released from the droplets.

Kinetics of Flavor Release. The taste of an emulsion depends on the rate at which the flavormolecules move from the food to the receptors on the tongue and inside of the mouth. Flavormolecules may be located in either the oil or water phase, although it is widely believed thattaste perception is principally a result of those molecules which are present in the water phase(McNulty 1987), because the flavor must cross an aqueous membrane before reaching thetaste receptors (Thomson 1986). An indication of the kinetics of flavor release can thereforebe obtained from a knowledge of the time dependence of the flavor concentration in theaqueous phase.

When an oil-in-water emulsion is diluted with saliva, some of the flavor molecules in thedroplets move into the aqueous phase. A mathematical model has been developed to describethe rate at which a solute is released from a spherical droplet surrounded by a finite volumeof a well-stirred liquid (Crank 1975):

M

M

Dq t r

qt n

nn∞ =

∞= −

+ −+ +∑1

6 1 1

9 9

2 2

2 21

α αα α

( ) exp( / )(9.19)

where Mt is the total amount of solute which has left the sphere by time t, M∞ is the totalamount of solute which has left the sphere once equilibrium has been established, D is thetranslational diffusion coefficient of the flavor within the droplets, t is the time, r is thedroplet radius, α = 3V/(4πr 3)KDC, V is the volume of the continuous phase, and qn are thenonzero roots of the relation, tan qn = 3qn/(1+αqn

2). This equation assumes that the concen-tration of solute (flavor) in the aqueous phase is initially zero, and therefore this equation isonly strictly applicable to emulsions which are diluted with high concentrations of saliva.Nevertheless, it does provide some useful insights into the rate of flavor release from oildroplets.

The influence of droplet radius on the flavor release rate from droplets in a typical oil-in-water emulsion is shown in Figure 9.7. The release rate increases as the size of the dropletsdecreases. For the system shown in Figure 9.7, the time required for half of the flavor to leavethe emulsion droplets is given by t1/2 = (0.162r)2/D. The variation of t1/2 with droplet radius

FIGURE 9.7 Kinetics of flavor release from spherical droplets suspended in a liquid.


TABLE 9.2Influence of Droplet Size on theTime It Takes for the Flavor toDiffuse Out of the Droplets

Radius (m) t 1/2 (s)

0.1 6.6 × 10–7

0.5 1.6 × 10–5

1 6.6 × 10–5

5 1.6 × 10–3

10 6.6 × 10–3

50 1.6 × 10–1

is shown in Table 9.2. Flavor release is extremely rapid in emulsions that contain droplets lessthan about 10 µm, but becomes appreciably slower at large sizes. It therefore seems that themovement of flavor molecules from the droplets to the aqueous phase will not be the rate-limiting step in flavor release provided the emulsion is well-agitated and the droplets arerelatively small. Instead, it is more likely that the breakdown of the structure and the diffusionof the molecules through the aqueous phase to the receptors are rate limiting. If the dropletsare coated by interfacial membranes, the release rate may be slowed considerably and wouldbe dependent on the type of emulsifier present, although little work has been carried out inthis area (Harvey et al. 1995).

9.2.2.2. Release of Volatile Compounds (Aroma)

The release of volatile compounds from a food involves the mass transfer of the compoundsthrough the emulsion and into the vapor phase (Overbosch et al. 1991). We are thereforeinterested in the change in the concentration of the flavor in the vapor phase with time.

Flavor Release from Homogeneous Liquids. The release rate of volatile flavors from ahomogeneous liquid (e.g., oil or water) has been modeled by assuming that static diffusionconditions apply (Overbosch et al. 1991). A finite thickness of the liquid containing the flavoris assumed to be in contact with an infinite gas phase. The flavor moves from the liquid intothe gas phase in order to try and establish the equilibrium partition coefficient (which cannever be achieved because of the infinite extent of the gas phase). The flux of the flavoracross the boundary separating the liquid and the gas and the change in the flavor concen-tration in the liquid with time are given by:

J Fc D tL L= 0 / π (9.20)

M t Fc D tL L( ) /= 2 0 π (9.21)

where

FK D D

K D D

GL G L

GL G L

=+

/

/1(9.22)

Here, DL and DG are the translational diffusion coefficients of the flavor in the liquid andthe gas phase, respectively; t is the time; KGL (= cG/cL) is the equilibrium partition coefficient


between the gas and liquid; and cL0 is the initial concentration of flavor in the liquid. The

factor F is the driving force for the diffusion process: its value varies from 0 (low flavorrelease rate) to 1 (high flavor release rate). The value of F is determined principally by theequilibrium partition coefficient of the flavor between the gas and the liquid (KGL), becausethere are much larger variations in this parameter than in diffusion coefficients for differentflavors.

These equations indicate that the rate of flavor release from a good solvent (KGL low) ismuch slower than the rate from a poor solvent (KGL high). Thus a nonpolar flavor will bereleased more slowly from a nonpolar solvent than a polar solvent and vice versa. Thisaccounts for the fact that the ranking of the release rates of flavors from water is opposite tothat from oil (Overbosch et al. 1991). It also indicates that anything which decreases thediffusion coefficient of the flavor molecules in the liquid will decrease the release rate (e.g.,increasing viscosity or molecular size).

Flavor release does not occur under static conditions, and therefore the above model haslimited applicability in practice. In reality, there will be a flow of gas over the food in themouth due to respiration and swallowing (Thomson 1986). When the flavor is constantlyswept away from the surface of the food, the concentration gradient is increased, and so therate of flavor loss is more rapid than under static conditions. Overbosch et al. (1991)developed a model to take this convective diffusion process into account. A numericalsolution of this model indicated that the release rate due to convective diffusion is muchgreater than that due to static diffusion when KGL is small, because the rate-limiting step isthe movement of the flavor compound from the liquid surface into the gas. At large KGL

values, the difference between the two theories is much smaller because the rate-limiting stepis the diffusion of the molecules through the liquid, rather than from the surface into the gas.Under convective conditions, the rate of flavor release therefore depends on the equilibriumpartition coefficient of the flavor, as well as the flow rate of the gas through the mouth andinto the nose.

Influence of Ingredient Interactions. A number of ingredients commonly found in foodemulsions are capable of decreasing the rate of flavor release because of their ability toeither bind flavors or retard their mass transfer (e.g., proteins, carbohydrates, and surfactantmicelles).

The effect of flavor binding on the rate of flavor release can be accounted for using thesame approach as for static diffusion in homogeneous liquids (i.e., Equations 9.20 to 9.22)(Overbosch et al. 1991). For reversible binding, DL is replaced by an effective diffusioncoefficient DL

e = DL/(K* +1), and KGL is replaced by an effective partition coefficient KeGL =

KGL/(K* +1), where K* is the binding coefficient (K* = cL,B/cL,F). The release rate is reducedbecause DL

e and KLe are smaller than DL and KGL, although the total amount of flavor released

when the process is allowed to go to completion is unchanged. For irreversible binding, cL0

is replaced by cL,F in Equations 9.20 to 9.22 because the rest of the flavor is “lost.” In addition,the total amount of flavor released when the process is allowed to go to completion is reducedby a factor cL,F/cL

0.The release rate may also be reduced because of the ability of certain food ingredients to

retard the movement of flavor molecules to the surface of the liquid, which may be due toan enhanced viscosity or due to structural hindrance (Kokini 1987). The diffusion coefficientof a molecule is inversely proportional to the viscosity of the surrounding liquid, and soincreasing the viscosity of the liquid will decrease the rate of flavor release because themovement of the flavor molecules is reduced. The presence of a network of aggregatedbiopolymer molecules may provide a physical barrier through which the molecules cannotdirectly pass. Instead, they may have to take a tortuous path through the network, whichincreases the time taken for them to reach the surface. If the flavor molecules are associated


with surfactant micelles, their release rate will depend on the diffusion coefficient of themicelles, as well as the kinetics of micelle breakdown (Section 4.5).

Highly volatile flavors (high KGL) are most affected by viscosity or structural hindranceeffects because the rate-limiting step in their release from a food is the movement throughthe liquid rather than the movement away from the liquid surface. On other hand, low-volatility flavors (low KGL) are affected less, because the rate-limiting step in their release isthe movement away from the liquid surface rather than through the liquid (Roberts et al.1996).

A great deal of research has been carried out to establish the relative importance of bindingand retarded mass transfer mechanisms. Many experimental studies have shown that increas-ing the biopolymer concentration decreases the rate of flavor release, but have been unableto establish the relative importance of the two mechanisms (Hau et al. 1996, Guichard 1996,Roberts et al. 1996). The importance of rheology has been demonstrated by studies whichhave shown that the intensity of flavors decreases as the viscosity of a biopolymer solutionor the strength of a biopolymer gel increases (Baines and Morris 1989, Carr et al. 1996). Onthe other hand, solutions with the same viscosity often have different flavor release rates,which may be because they have different microstructures or because the flavors bind to themdifferently (Guichard 1996). It is clear that more systematic research is needed to establishthe role of biopolymers and other ingredients in retarding flavor release. The influence ofingredient interactions on release rates has important consequences for the formulation ofmany food products. For example, it may be necessary to incorporate more flavor into a foodto achieve the same flavor intensity when the biopolymer concentration is increased.

Flavor Release from Emulsions. Overbosch et al. (1991) used the same mathematical ap-proach as for homogeneous liquids to describe the rate of flavor release from emulsions. Forstatic diffusion conditions, the flux and time dependence of the mass of flavor in an emulsionare given by:

J F c D tE Eo

E= / π (9.23)

M t Fc D tEo

E( ) /= 2 π (9.24)

where

FK D D

K D Dc c cE

GE G E

GE G E

E C D=+

= − +/

/( )

11 φ φ

DD D

KK

c

c

K K

KEC D

DCGE

G

E

GD DC

DC

=− ++ −

= =+ −

( )

( ) ( )

1

1 1 1 1

φ φφ φ

These equations provide some useful insights into the major factors which determine flavorrelease in emulsions. They indicate that the release rate depends on the diffusion and partitioncoefficients of the flavor in the oil, water, and gas phases. The larger the value of KGE or DE,the more rapid the flavor release. Similar conclusions were recently drawn by Harrison et al.(1997) using another mathematical model based on penetration theory.

The above equations predict that the release rate from an oil-in-water emulsion is the sameas that from a water-in-oil emulsion with the same composition, because they are symmetrical


with respect to the physical properties of the two phases (Overbosch et al. 1991). This willonly be true if the flavor in the droplets and continuous phase is in equilibrium. Some studieshave indicated that the taste perception of oil-in-water and water-in-oil emulsions of the samecomposition is approximately the same (Barylko-Pikielna et al. 1994, Brossard et al. 1996).Nevertheless, other studies have shown that the release rate is faster from oil-in-wateremulsions than from water-in-oil emulsions, which suggests that the emulsion cannot besimply treated as a homogeneous liquid with “averaged” properties (Overbosch et al. 1991,Bakker and Mela 1996).

The above equations are most likely to be suitable for describing systems in which the rate-limiting step is the movement of the flavor molecules from the emulsion surface to the gasphase (i.e., when KGE is small), rather than those where the movement of the flavor moleculesthrough the emulsion is rate limiting (i.e., when KGE is large), especially for emulsions withrelatively large droplet sizes.

It is possible to develop more sophisticated theories to describe the kinetics of flavorrelease from emulsions which take into account the partitioning and diffusion of the flavormolecules in the oil, water, gas, and interfacial membrane. Nevertheless, these theories aremuch more complex and can usually only be solved numerically.

9.2.3. Measurements of Partition Coefficientsand Flavor Release

A variety of experimental techniques can be used to measure equilibrium partition coeffi-cients and the kinetics of flavor release in emulsions.

9.2.3.1. Head Space Analysis

The concentration of flavor in the vapor phase above a liquid can be determined by headspace analysis (Franzen and Kinsella 1975, Overbosch et al. 1991, O’Neill 1996, Landy etal. 1996, Guyot et al. 1996). The liquid is placed in a sealed container which is stored underconditions of constant temperature and pressure. Samples of the gas phase are removed fromthe head space using a syringe which is inserted through the lid of the sealed container, andthe concentration of flavor is measured, usually by gas chromatography or high-performanceliquid chromatography. The concentration of volatile flavors in many foods is too low to bedetected directly by conventional chromatography techniques, and therefore it is necessary toconcentrate the samples prior to analysis. Head space analysis can be carried out over timeto monitor the kinetics of flavor release or after the sample has been left long enough forequilibrium to be attained to determine equilibrium partition coefficients.

9.2.3.2. Concentration Analysis in Static Binary Liquids

The equilibrium partition coefficient of a flavor between a bulk oil and bulk water phase canbe determined by measuring the concentration of flavor in the two phases after the systemhas been left long enough to attain equilibrium (McNulty 1987, Guyot et al. 1996, Huanget al. 1997). The flavor is usually added to one of the liquids first, and then the water phaseis poured into the container and the oil phase is poured on top. The container is sealed andstored in a temperature-controlled environment until equilibrium is achieved, which can bea considerable period (a few days or weeks), although this time can be shortened by mildagitation of the sample. The method used to determine the concentration depends on thenature of the flavor molecule. The most commonly used techniques are spectrophotometry,chromatography, and radiolabeling. In some systems, it is possible to analyze the solutionsdirectly, whereas in others it is necessary to extract the flavors first using appropriatesolvents.


The partition coefficient of flavors in emulsions can be determined using a similar proce-dure. The emulsion to be analyzed is placed into a container and a known amount of the flavoris added to the continuous phase. The container is filled to the top and sealed to prevent anyof the flavor from partitioning into the vapor phase. It is then stored in a temperature-controlled environment until it reaches equilibrium. Equilibrium is attained much morerapidly than in a nonemulsified system because the molecules only have to diffuse a shortdistance through the droplets. The emulsion is centrifuged to separate the droplets from thecontinuous phase, and then a sample of the continuous phase is removed for analysis of theflavor concentration. The partition coefficient can then be determined from a knowledge ofthe overall flavor concentration: KDC = cD/cC = (ctotal – cC)/cC. One important limitation of thistechnique is that it cannot distinguish between the flavor which is contained within thedroplets and that which is associated with the interfacial membrane.

9.2.3.3. Concentration Analysis in Stirred Diffusion Cells

Flavor release occurs under highly dynamic conditions within the mouth (Land 1996), andso a number of workers have developed experimental techniques which attempt to mimicthese conditions. McNulty and Karel (1973) developed a stirred diffusion cell for monitoringthe mass transport of flavor compounds between an oil and aqueous phase under shearconditions. The flavor compound is initially dissolved in either the oil or aqueous phase. Aknown volume of the aqueous phase is then poured into the vessel, and a known volume ofoil is poured on top. The oil and aqueous phases are sheared separately using a pair of stirrers,and samples are extracted periodically using syringes which protrude into each of the liquids(Figure 9.8). These samples are then analyzed to determine the concentration of the flavorwithin them. The liquids are stirred at a rate which ensures a uniform flavor distribution,without significantly disturbing the air–water or oil–water interfaces. By carrying out themeasurements over a function of time, it is possible to determine the kinetics of flavortransport between the oil and water.

A similar system can be used to study the movement of flavor from a solution or emulsionto the vapor phase above it. The liquid to be analyzed is placed in a sealed vessel and stirredat a constant rate. A syringe is used to withdraw samples from the head space above the liquidas a function of time. Alternatively, a continuous flow of gas can be passed across the stirredliquid and the concentration of flavor within it determined by chromatography (Roberts andAcree 1996). Thus it is possible to simulate the agitation of the food within the mouth, as wellas the flow of the gas across the food during manufacture.

FIGURE 9.8 Diagram of stirred diffusion cells used to measure the kinetics of flavor release in liquidsand emulsions.


9.2.3.4. Sensory Analysis

The ultimate test of the flavor profile of a food is its acceptance by consumers. Analyticaltests carried out in a laboratory help to identify the most important factors which determineflavor release, but they cannot model the extreme complexity of the human sensory system.For this reason, many researchers use sensory analysis by human subjects to assess the overallflavor profile of a food sample (Buttery et al. 1973, Williams 1986, Barylko-Pikielna et al.1994, Guyot et al. 1996).

9.3. EMULSION APPEARANCE

The first impression that a consumer usually has of a food emulsion is a result of itsappearance (Francis and Clydesdale 1975, Hutchings 1994). Appearance therefore plays animportant role in determining whether or not a consumer will purchase a particular product,as well as his or her perception of the quality once the product is consumed. A number ofdifferent characteristics contribute to the overall appearance of a food emulsion, including itsopacity, color, and homogeneity. These characteristics are the result of interactions betweenlight waves and the emulsion. The light which is incident upon an emulsion may be reflected,transmitted, scattered, absorbed, and refracted before being detected by the human eye(Francis and Clydesdale 1975, Farinato and Rowell 1983, Hutchings 1994, Francis 1995). Abetter understanding of the relationship between the appearance of emulsions and theircomposition and microstructure will aid in the design of foods with improved quality. Thissection highlights some of the most important factors which contribute to the overall appear-ance of emulsions.

9.3.1. Interaction of Light Waves with Emulsions

9.3.1.1. Transmission, Reflection, and Refraction

When an electromagnetic wave is incident upon a boundary between two homogeneousnonabsorbing materials, it is partly reflected and partly transmitted (or refracted) (Figure 9.9).The relative importance of these processes is determined by the refractive indices of the twomaterials, the surface topography, and the angle at which the light meets the surface (Hutchings1994).

The reflection of an electromagnetic wave from a surface may be either specular or diffuse.Specular reflectance occurs when the angle of reflection is equal to the angle of incidence(φreflection = φincidence) and is the predominant form of reflection from optically smooth surfaces.

FIGURE 9.9 When a light wave encounters a planar interface between two materials, it is partlyreflected and partly transmitted.


FIGURE 9.10 Comparison of specular and diffuse reflectance.

Diffuse reflectance occurs when the light is reflected over many different angles and is themost important form of reflection from optically rough surfaces (Figure 9.10).

When the angle of incidence is perpendicular to the surface of a material, the fraction oflight which is specularly reflected is given by the reflection coefficient (R)

Rnn

= −+

11

2

(9.25)

where n is the relative refractive index (= n2/n1), and n1 and n2 are the refractive indices ofthe two materials (Hutchings 1994). The greater the difference in refractive index betweenthe two materials, the larger the fraction of light which is reflected. About 0.1% of light isreflected from an interface between oil (n0 = 1.43) and water (nW = 1.33), while about 2% isreflected from an interface between water and air (nA = 1). These reflection coefficients mayseem small, but the total amount of energy reflected from a concentrated emulsion becomessignificant because a light wave encounters a huge number of different droplets and isreflected from each one of them.

When a light wave encounters a smooth material at an angle, part of the wave is reflectedat an angle equal to that of the incident wave, while the rest is refracted (transmitted) at anangle which is determined by the relative refractive indices of the two materials and the angleof incidence: sin(φrefraction) = sin(φincidence)/n.

9.3.1.2. Absorption

Absorption is the process whereby a photon of electromagnetic energy is transferred to anatom or molecule (Atkins 1994, Penner 1994a, Pomeranz and Meloan 1994). The primarycause of absorption of electromagnetic radiation in the visible region is the transition ofouter-shell electrons from lower to higher electronic energy levels. A photon is only ab-sorbed when it has an energy which exactly corresponds to the difference between theenergy levels involved in the transition, that is ∆E = hv = hc/λ, where h is Planck’s constant,v is the frequency of the electromagnetic wave, c is the velocity of the wave, and λ is thewavelength.

The visible region consists of electromagnetic radiation with wavelengths between about380 and 750 nm, which corresponds to energies of between about 120 and 230 kJ mol–1 inwater (Penner 1994a). Substances which can absorb electromagnetic energy in this region areusually referred to as chromophores (Patterson 1967). The most common type of chro-mophoric groups which are present in food emulsions are attached to organic moleculeswhich contain conjugated unsaturated bonds or aromatic ring structures (Hutchings 1994).Single unsaturated bonds tend to absorb in the ultraviolet rather than the visible region(Pomeranz and Meloan 1994).


Absorption causes a reduction in the intensity of a light wave as is passes through amaterial, which can be described by the following equation (Penner 1994b, Pomeranz andMeloan 1994):

TI

IS=0

(9.26)

where T is the transmittance, IS is the intensity of the light which travels directly through thesample, and I0 is the intensity of the incident wave. In practice, IS is also reduced because ofreflections from the surfaces of the cell and due to absorption by the solvent and cell (Penner1994b). These losses can be taken into account by comparing IS with the intensity of a wavewhich has traveled through a reference cell (which usually contains pure solvent), rather thanwith I0:

TI

IS

R

= (9.27)

where IR is the intensity of the light which has traveled directly through the reference cell.The transmittance of a substance decreases exponentially with increasing chromophore con-centration or sample length (Pomeranz and Meloan 1994). For this reason, it is often moreconvenient to express the absorption of light in terms of an absorbance (A) because this isproportional to the chromophore concentration and the sample length:

AI

IclS

R

= =log α (9.28)

where α is a constant of proportionality known as the absorptivity, c is the chromophoreconcentration, and l is the sample path length. The linear relationship between absorbanceand concentration holds over the chromophore concentrations used in most food emulsions.The color intensity of a particular chromophore therefore depends on its concentration andabsorptivity.

The appearance of an emulsion to the human eye is determined by the interactions betweenit and electromagnetic radiation in the visible region (Francis and Clydesdale 1975, Hutchings1994). It is therefore important to measure the absorbance over the whole range of visiblewavelengths (390 to 750 nm). A plot of absorbance versus wavelength is referred to as anabsorption spectrum (Figure 9.11). An absorption peak occurs at a wavelength which de-pends on the difference between the energy levels of the electronic transitions in the chro-mophores (λ = ch/∆E). Absorption peaks are fairly broad in the visible region becausetransitions occur between the different vibrational and rotational energy levels within theelectronic energy levels and because of interactions between neighboring molecules (Penner1994b).

9.3.1.3. Scattering

Scattering is the process whereby a wave which is incident upon a particle is directed intodirections which are different from that of the incident wave (Farinato and Rowell 1983,Hiemenz 1986). The extent of light scattering by an emulsion is determined mainly by therelationship between the droplet size and wavelength and by the difference in the refractiveindex between the droplets and the surrounding liquid.


FIGURE 9.11 Absorption spectrum of colored liquids.

Light scattering causes a reduction in the intensity of an electromagnetic wave as it passesthrough an emulsion, which is characterized by the transmittance (Farinato and Rowell1983):

TI

IlS

R

= = −exp( )τ (9.29)

where τ is the turbidity of the emulsion. In dilute emulsions (<0.05%), the turbidity is propor-tional to the droplet concentration, but in concentrated emulsions, the turbidity falls below theexpected value because of multiple scattering effects (Ma et al. 1990).

Turbidity is a desirable attribute of many food emulsions because it imparts a natural-looking character and appeal (Hernandez and Baker 1991, Hernandez et al. 1991). Forexample, low concentrations of oil droplets consisting of vegetable or flavor oils are used toprovide a turbid appearance to many types of fruit beverage (Tan 1990, Dickinson 1994). Theturbidity of these emulsions depends principally on the size, concentration, and relativerefractive index of the droplets.

The turbidity of an emulsion is related to the characteristics of the droplets which itcontains by the following relationship (Hernandez and Baker 1991):

τ πρ

=

34 1

c Q

r (9.30)

where c is the droplet concentration, ρ1 is the density of the continuous phase, r is the dropletradius, and Q is the scattering cross-section of the individual particles:

Q = − + −24 4

12β

ββ

βsin ( cos ) (9.31)

where

α πλ

β α= = −22 1

rnn


FIGURE 9.12 The turbidity of an emulsion depends on the droplet size and wavelength.

These equations have to be modified for polydisperse emulsions to take into accountthe fact that each droplet has its own scattering profile (Hernandez and Baker 1991). Itshould be noted that Equation 9.31 is only applicable to emulsions in which the relativerefractive index is close to unity and the size of the droplets is greater than the wavelengthof light.

The variation of emulsion turbidity with droplet size and wavelength is illustrated in Figure9.12. The dependence of scattering on droplet size can be conveniently divided into threeregions: long-wavelength regime (r << λ), intermediate-wavelength regime (r ≈ λ), and short-wavelength regime (r >> λ). The turbidity is greatest in the intermediate-wavelength regime,where the size of the droplets is similar to that of the wavelength. The wavelength of lightvaries from about 0.3 to 0.7 µm, and therefore droplets of this size would be expected toscatter most strongly.

Experiments with citrus oil-in-water emulsions have shown that there is a maximum intheir turbidity (at 650 nm) when the droplets have a diameter of just under 1 µm (Hernandezand Baker 1991). This has important implications for the formulation of food beveragesbecause it means that the characteristic turbidity of a product can be obtained using a lowerconcentration of oil when the droplet diameter is optimized to give the maximum amount ofscattering. Nevertheless, other factors which depend on particle size also have to be consid-ered, such as the stability of the product to creaming or sedimentation.

The influence of the relative refractive index on the scattering cross-section of an emul-sion is illustrated in Figure 9.13. The scattering is minimum when the droplets have arefractive index equal to the surrounding liquid, but increases as the refractive index movesto higher or lower values. The refractive index of pure water is 1.33 (Walstra 1968), whilethat of most food oils ranges between 1.4 and 1.5 (Formo 1979).* The refractive indexof aqueous solutions depends on the type and concentration of solutes present (e.g., pro-teins, sugars, alcohols, and salts) (Figure 9.14). At high concentrations of these compo-nents, it is possible for the refractive index of the droplets to be matched to that of thesurrounding liquid, which greatly reduces the degree of scattering by the droplets andtherefore causes the emulsion to appear transparent (Taisne et al. 1996). This effect canclearly be observed by diluting an oil-in-water emulsion in a series of aqueous solutions thatcontain different concentrations of sucrose. An emulsion which appears optically opaquewhen the droplets are dispersed in water becomes transparent when they are dispersed in aconcentrated sugar solution (≈60%), even though the droplet size and concentration areunchanged.

* It should be noted that the refractive indices of substances vary with wavelength (Walstra 1968).


FIGURE 9.14 Dependence of the refractive index of various aqueous solutions on solute concentration.

9.3.1.4. Absorption and Scattering in Concentrated Emulsions

The optical characteristics of concentrated emulsions can be described by the Kubelka–Munktheory (Francis and Clydesdale 1975, Hutchings 1994). This theory enables one to determinethe relative amounts of scattering and absorption in a concentrated emulsion from measure-ments of the light reflected from it:

FKS

R

RKM = =− ∞

∞

( )1

2

2

(9.32)

where FKM is the Kubelka–Munk parameter, K and S are the absorption and scatteringcoefficients, and R∞ is the reflectance from an infinitely thick sample at a given wavelength.When an emulsion is highly colored, the absorption of light dominates (high FKM) and thereflectance is low. As scattering effects become more important (low FKM), the reflectanceincreases and the emulsion appears “lighter” in color. The scattering coefficient is propor-tional to the droplet concentration, while the absorption coefficient is proportional to thechromophore concentration. The appearance of an emulsion therefore depends on the concen-tration of the various components within it.

FIGURE 9.13 Influence of refractive index on the degree of scattering from emulsions. The scatteringis smallest in the region where the droplet and continuous phase have similar refractive indices, butincreases at lower or higher values.


The Kubelka–Munk theory can be used to predict the influence of different types ofscatterers or chromophores on the overall appearance of an emulsion. It can also be relatedto the scattering cross-sections of the individual droplets in an emulsion using diffuse scat-tering theory. It therefore provides a valuable mathematical link between the physicochemicalcharacteristics of the droplets and the appearance of an emulsion.*

9.3.2. Overall Appearance

The overall appearance of an emulsion is a result of the various types of interactions betweenlight waves and foods mentioned in the previous section.

9.3.2.1. Opacity

An object which allows all of the light to pass through it is referred to as being transparent,whereas an object which scatters or absorbs all of the light is referred to as being opaque(Clydesdale 1975). Many dilute emulsions fall somewhere between these two extremes andare therefore referred to as being translucent. The opacity of most food emulsions is deter-mined mainly by the scattering of light from the droplets: the greater the scattering, thegreater the opacity (Hernandez and Baker 1991, Dickinson 1994). The extent of scattering isdetermined by the concentration, size, and relative refractive index of the droplets (Section9.3.1.3). An emulsion becomes more opaque as the droplet concentration or scattering cross-section increases (Equation 9.29).

The scattering of light accounts for the opacity of emulsions that are made from two liquidswhich are themselves optically transparent (e.g., water and a mineral oil). When a light waveimpinges on an emulsion, all of the different wavelengths are scattered by the droplets, andso the light cannot penetrate very far into the emulsion. As a consequence, the emulsionappears to be optically opaque (Farinato and Rowell 1983).

9.3.2.2. Color

It is extremely difficult for human beings to objectively describe the colors of materials usingeveryday language (Hutchings 1994). For this reason, a number of standardized methodshave been developed to measure and specify color in a consistent way (Francis and Clydes-dale 1975, Hutchings 1994). The underlying principle of these methods is that all colors canbe simulated by combining three selected colored lights (red, green, and blue) in the appro-priate ratio and intensities. This trichromatic principle means that it is possible to describe anycolor in terms of just three mathematical variables (e.g., hue, value, and chroma) (Francis andClydesdale 1975). Experimental techniques for measuring these variables are discussedbriefly in Section 9.3.3.

The color of an emulsion is determined by the absorption and scattering of light wavesfrom both the droplets and continuous phase (Dickinson 1994). The absorption of lightdepends on the type and concentration of chromophores present, while the scattering of lightdepends on the size, concentration, and relative refractive index of any particulate matter.Whether an emulsion appears “red,” “orange,” “yellow,” “blue,” etc. depends principally onits absorption spectra. Under normal viewing conditions, an emulsion is exposed to whitelight from all directions.** When this light is reflected, transmitted, or scattered by the

* Recently, it has been shown how the color of an emulsion can be predicted using the Kubelka–Munk theory froma knowledge of the droplet and dye characteristics (McClements et al. 1998a).

** If an emulsion is placed in a dark room and a white light beam is directed through it, it appears blue whenobserved from the side and red when observed from the back, because blue light has a lower wavelength thanred light and is therefore scattered to a wider angle (Farinato and Rowell 1983).


emulsion, some of the wavelengths are absorbed by the chromophores present. The color ofthe light which reaches the eye is a result of the nonabsorbed wavelengths (e.g., an emulsionappears red if it absorbs all of the other colors except the red) (Francis and Clydesdale 1975).The color of an emulsion is modified by the presence of the droplets or any other particulatematter. As the concentration or scattering cross-section of the particles increases, an emulsionbecomes lighter in appearance because the scattered light does not travel very far through theemulsion and is therefore absorbed less by the chromophores. It is therefore possible tomodify the color of an emulsion by altering the characteristics of the emulsion droplets orother particulate matter.

9.3.2.3. Homogeneity

The quality of a food emulsion is often determined by the uniformity of its appearance overthe whole of its surface. An emulsion may have a heterogeneous appearance for a number ofreasons: (1) it contains particles which are large enough to be resolved by the human eye(>0.1 mm) or (2) it contains particles which have moved to either the top or the bottom ofthe container because of gravitational separation. Methods of retarding creaming and sedi-mentation were considered in Chapter 7.

9.3.3. Experimental Techniques

9.3.3.1. Spectrophotometry

A variety of different types of spectrophotometers have been developed to measure thetransmission and reflection of light from objects as a function of wavelength in the visibleregion (Clydesdale 1975, Francis and Clydesdale 1975, Pomeranz and Meloan 1994, Hutchings1994). These instruments usually consist of a light source, a wavelength selector, a sampleholder, and a light detector (Figure 9.15).

Transmission Spectrophotometer. A beam of white light, which contains electromagneticradiation across the whole of the visible spectrum, is passed through a wavelength selector,which isolates radiation of a specific wavelength (Penner 1994b). This monochromatic waveis then passed through a cell containing the sample, and the intensity of the transmitted waveis measured using a light detector. By comparing the intensity of the light transmitted by thesample with that transmitted by a reference material, it is possible to determine the transmit-tance of the sample (Equation 9.27). A transmittance spectrum is obtained by carrying out thisprocedure across the whole range of wavelengths in the visible region. Transmission mea-surements can only be carried out on emulsions which allow light to pass through, andtherefore they cannot be used to analyze concentrated emulsions.

Reflection Spectrophotometer. In these instruments, the intensity of light reflected from thesurface of a sample is measured (Francis and Clydesdale 1975). The reflectance (R) of amaterial is defined as the ratio of the intensity of the light reflected from the sample (RS) tothe intensity of the light reflected from a reference material of known reflectance (RR): R =RS/RR. The precise nature of the experimental device depends on whether the reflection isspecular or diffuse. For specular reflection, the intensity of the reflected light is usuallymeasured at an angle of 90° to the incident wave, whereas for diffuse reflection, the sum ofthe intensity of the reflected light over all angles is measured using a device called anintegrating sphere (Figure 9.15). A reflectance spectrum is obtained by carrying out thisprocedure across the whole range of wavelengths in the visible region.

The transmittance and reflectance spectra obtained from a sample can be used to calculatethe relative magnitudes of the absorption and scattering of light by an emulsion as a functionof wavelength. Alternatively, the color of a product can be specified in terms of trichromatic


coordinates by analyzing the spectra using appropriate mathematical techniques (McClementset al. 1998). The details of these techniques have been described elsewhere and are beyondthe scope of this book (Francis and Clydesdale 1975, Hutchings 1994).

9.3.3.2. Light Scattering

Light-scattering techniques are used principally to determine the size distribution of thedroplets in an emulsion (Chapter 10). A knowledge of the droplet size distribution enablesone to predict the influence of the droplets on light scattering and therefore on the turbidityof an emulsion (Hernandez and Baker 1991, Dickinson 1994). Alternatively, the experimen-tally determined scattering pattern — the intensity of scattered light versus scattering angle— can be used directly to describe the scattering characteristics of the emulsion droplets.

The scattering of light from dilute emulsions is sometimes characterized by a device knownas a nephelometer (Hernandez et al. 1991). This device measures the intensity of light whichis scattered at an angle of 90° to the incident beam. The intensity of light scattered by asample is compared with that scattered by a standard material of known scattering character-istics (e.g., formazin) (Hernandez et al. 1991). Because small droplets scatter light morestrongly at wide angles than large droplets, the nephelometer is more sensitive to the presenceof small droplets than are turbidity measurements.

9.3.3.3. Colorimetry

A large number of instruments have been developed to characterize the color of materialswhich are based on the trichromatic principle mentioned previously (Francis and Clydesdale1975, Hutchings 1994, Francis 1995). A simple colorimeter consists of a light source, thesample to be analyzed, a set of three filters (red, green, and blue), and a photocell to determinethe light intensity (Figure 9.16). Colorimetry measurements can be carried out in either atransmission or a reflection mode. In the transmission mode, the intensity of a light beam ismeasured after it has been passed through the sample and one of the color filters. This

FIGURE 9.15 Examples of transmission and reflection spectrophotometers.


FIGURE 9.16 Schematic diagram of a simple colorimeter.

procedure is then repeated for the other two filters, and the sample is characterized by theintensity of the light wave passing through the red, green, and blue filters. In the reflectionmode, the light beam is passed through a filter and is then reflected from the surface of asample. For specular reflection, a photocell is usually positioned at 90° to the incident wave,while for diffuse reflection, an integrating sphere is used to determine the intensity of thereflected light. This procedure is carried out for each of the three color filters, so that the colorof the sample is characterized in terms of the intensities of the red, green, and blue lights.Reflectance measurements are most suitable for determining the color of concentrated emul-sions, while transmission measurements are more suitable for characterizing dilute emulsions.

9.3.3.4. Sensory Analysis

Ultimately, the appearance of an emulsion must be acceptable to the consumer, and thereforeit is important to carry out sensory tests (Hutchings 1994). These tests can be carried out usingeither trained specialists or untrained individuals. Each individual is given one or moresamples to analyze and asked either to compare the appearance of the samples with each otheror to rank certain attributes of the appearance of each sample. Sensory tests must be carriedout in a room where the light source is carefully controlled to obtain reproducible measure-ments which correspond to the conditions a person might experience when consuming theproduct (Francis and Clydesdale 1975).


10 Characterization ofEmulsion Properties

10.1. INTRODUCTION

Previous chapters stressed the relationship between the bulk physicochemical characteristicsof food emulsions and their colloidal properties. The more efficient production of high-quality food products depends on an improved understanding of this relationship, as well ason the successful implementation of this knowledge in practice. Advances in these areas relyon the availability of analytical techniques to characterize emulsion properties. These tech-niques are needed in the laboratory for research and development and in the food factory tomonitor the properties of foods before, during, and after production.

10.1.1. Research and Development

The properties of foods can be considered at a number of different levels, ranging from themolecular to the bulk physicochemical and organoleptic characteristics of the final product(Baianu 1992, Eads 1994). The quality attributes of a food emulsion are therefore the resultof a complex set of interactions between and within the different levels of structural organi-zation (e.g., molecular, colloidal, microscopic, macroscopic, and organoleptic). This relation-ship is made even more complex because of the dynamic nature of food emulsions. Thechemical structure or spatial organization of the components within an emulsion may changewith time or in response to changes in environmental conditions. A major concern of foodscientists is identifying the most important factors which determine the properties at eachlevel and determining the relationship between the different levels. To elucidate these rela-tionships, it is necessary to use a combination of experimental techniques and theoreticalunderstanding. An array of different experimental techniques are available for studyingemulsion properties, ranging from the molecular to the organoleptic. By using a combinationof these techniques, it is possible to obtain a greater understanding of the relationship betweenthe different levels of structural organization. This type of information can then be used tosystematically design and manufacture foods with improved properties.

10.1.2. Quality Control

The information obtained from research and development is used to select the most appro-priate ingredients and processing operations required to manufacture a product with thedesired quality attributes (Kokini et al. 1993). Nevertheless, there are always inherent varia-tions in the raw materials (e.g., composition, quality) and processing conditions (e.g., tem-peratures, pressures, times, flow rates) used to manufacture food products, which inevitablyleads to variability in the properties of the final product. To minimize these variations, it isnecessary to understand the contribution that each ingredient makes to the overall propertiesand the influence of processing conditions on these properties. This type of informationcomes from the fundamental studies discussed in Section 10.1.1.


* The emulsifying capacity of an oil-soluble emulsifier can be determined in the same way, except that the wateris titrated into the oil phase.

Analytical techniques are also needed to monitor the properties of foods at each stage ofmanufacturing (Nielsen 1994). The data obtained from these techniques are used to optimizeprocessing conditions, thus improving product quality, reducing waste, improving energyefficiency, and minimizing variations in the properties of the final product.

As mentioned earlier, the overall quality of food emulsions is determined by a wide rangeof different factors, ranging from the molecular to the organoleptic. In this chapter, the focusis on those experimental techniques which measure properties that are unique to emulsions(i.e., emulsifier efficiency, dispersed-phase volume fraction, droplet size distribution, dropletcrystallinity, and droplet charge). In previous chapters, experimental techniques were re-viewed for characterizing interfacial properties (Chapter 5), emulsion stability (Chapter 7),and emulsion rheology (Chapter 8). Despite their importance in understanding the propertiesof food emulsions, we have not considered techniques for measuring molecular, chemical,enzymatic, microbiological, or organoleptic properties, because these techniques are commonto all foods and are therefore beyond the scope of this book.

10.2. TESTING EMULSIFIER EFFICIENCY

One of the most important decisions a food manufacturer must make when developing anemulsion-based food product is the selection of the most appropriate emulsifier (Fisher andParker 1985, Charalambous and Doxastakis 1989, Dickinson 1992, Hasenhuettl 1997). Ahuge number of emulsifiers are available as food ingredients, and each has its own uniquecharacteristics and optimum range of applications (Hasenhuettl and Hartel 1997). The effi-ciency of an emulsifier is governed by a number of characteristics, including the minimumamount required to produce a stable emulsion, its ability to prevent droplets from aggregatingover time, the speed at which it adsorbs to the droplet surface during homogenization, theinterfacial tension, and the thickness and viscoelasticity of the interfacial membrane. Thesecharacteristics depend on the food in which the emulsifier is present and the prevailingenvironmental conditions (e.g., pH, ionic strength, ion type, oil type, ingredient interactions,temperature, and mechanical agitation) (Sherman 1995). For this reason, it is difficult toaccurately predict the behavior of an emulsifier from a knowledge of its chemical structure(although some general prediction about its functional properties is usually possible). Instead,it is often better to test the efficiency of an emulsifier under conditions which are similar tothose found in the actual food product in which it is going to be used (Sherman 1995). Anumber of procedures commonly used to test emulsifier efficiency are discussed in thissection.

10.2.1. Emulsifying Capacity

It is often important for a food manufacturer to know the minimum amount of an emulsifierthat can be used to create a stable emulsion. The emulsifying capacity of a water-solubleemulsifier is defined as the maximum amount of oil that can be dispersed in an aqueoussolution that contains a specific amount of the emulsifier without the emulsion breaking downor inverting into a water-in-oil emulsion (Sherman 1995). Experimentally, it is determined byplacing an aqueous emulsifier solution into a vessel and continuously agitating using a high-speed blender as small volumes of oil are titrated into the vessel (Swift et al. 1961, Das andKinsella 1990).* The end point of the titration occurs when the emulsion breaks down orinverts, which can be determined by optical, rheological, or electrical conductivity measure-ments. The greater the volume of oil which can be incorporated into the emulsion before it


breaks down, the higher the emulsifying capacity of the emulsifier. Although this test iswidely used to characterize emulsifiers, it has a number of drawbacks which limit its appli-cation as a standard procedure (Sherman 1995, Dalgleish 1996a). The main problem with thetechnique is that the amount of emulsifier required to stabilize the emulsion is governed bythe oil–water interfacial area rather than by the oil concentration, and so the emulsifyingcapacity depends on the size of the droplets produced during agitation. As a consequence, theresults are particularly sensitive to the type of blender and blending conditions used in thetest. In addition, the results of the test have also been found to depend on the rate at whichthe oil is titrated into the vessel, the method used to determine the end point, the initialemulsifier concentration, and the measurement temperature (Sherman 1995). The emulsify-ing capacity should therefore be regarded as a qualitative index which depends on the specificconditions used to carry out the test. Nevertheless, it is useful for comparing the efficiencyof different emulsifiers under the same experimental conditions.

A more reliable means of characterizing the minimum amount of emulsifier required toform an emulsion is to measure the surface load (ΓS), which corresponds to the mass ofemulsifier required to cover a unit area of droplet surface (Dickinson 1992). A stable emul-sion is prepared by homogenizing known amounts of oil, water, and emulsifier. The mass ofemulsifier adsorbed to the surface of the droplets per unit volume of emulsion (Ca/kg m–3)is equal to the initial emulsifier concentration minus that remaining in the aqueous phase afterhomogenization (which is determined by centrifuging the emulsion to remove the dropletsand then analyzing the emulsifier concentration in the serum). The total droplet surface areacovered by the adsorbed emulsifier is given by S = 6φVe/d32, where Ve is the emulsion volumeand d32 is the volume–surface mean droplet diameter. Thus the surface load can be calculated:ΓS = CaVe/S = Cad32/6φ, which is typically a few milligrams per meter squared. A knowledgeof the surface load enables one to calculate the minimum amount of emulsifier required toprepare an emulsion that contain droplets of a given size and concentration. In practice, anexcess of emulsifier is usually needed because it does not all adsorb to the surface of thedroplets during homogenization due to the finite time it takes for an emulsifier to reach theoil–water interface and because there is an equilibrium between the emulsifier at the dropletsurface and that in the continuous phase (Hunt and Dalgleish 1994, Dalgleish 1996a). Inaddition, the surface load is often dependent on environmental conditions, such as pH, ionicstrength, temperature, and protein concentration (Dickinson 1992, Hunt and Dalgleish 1994,Dalgleish 1996a).

10.2.2. Emulsion Stability Index

An efficient emulsifier produces an emulsion in which there is no visible separation of theoil and water phases over time. Phase separation may not become visible to the human eyefor a long time, even though some emulsion breakdown has occurred. Consequently, it isimportant to have analytical tests which can be used to detect the initial stages of emulsionbreakdown, so that their long-term stability can be predicted.

One widely used test is to centrifuge an emulsion at a given speed and time and observethe amount of creaming and/or oil separation which occurs (Smith and Mitchell 1976,Tornberg and Hermannson 1977, Aoki et al. 1984, Das and Kinsella 1990). This test can beused to predict the stability of an emulsion to creaming using relatively low centrifuge speedsor to coalescence by using speeds which are high enough to rupture the interfacial mem-branes. The greater the degree of creaming or oil separation that occurs, the greater theinstability of an emulsion and the less efficient the emulsifier.

An alternative approach which can be used to accelerate emulsion instability is to measurethe degree of droplet coalescence when an emulsion is subjected to mechanical agitation(Britten and Giroux 1991, Dickinson et al. 1993b, Dickinson and Williams 1994). The droplet


size distribution of the emulsions can be measured either as a function of time as theemulsions are agitated at a constant stirring speed or as a function of stirring speed after theemulsions have been agitated for a fixed time. The faster the increase in droplet size withtime, the greater the instability of the emulsion and the lower the efficiency of the emulsifier.

Although these tests are widely used and can be carried out fairly rapidly, they do have anumber of important limitations: the rate of creaming or coalescence in a centrifugal field orduring agitation may not be a good indication of emulsion instability under normal storageconditions, and it does not take into account chemical or biochemical reactions that mightalter emulsion stability over extended periods.

A more quantitative method of determining emulsifier efficiency is to measure the changein the particle size distribution of an emulsion with time using one of the analytical techniquesdiscussed in Section 10.3. An efficient emulsifier produces emulsions in which the particlesize distribution does not change over time, whereas a poor emulsifier produces emulsionsin which the particle size increases due to coalescence and/or flocculation. The kinetics ofemulsion stability can be established by measuring the rate at which the particle size increaseswith time. These tests should be carried out under conditions similar to those found in thefinal product (e.g., pH, ionic strength, composition, temperature, etc.).

The analytical instruments used for measuring particle size distributions are often fairlyexpensive and are not available in many small institutions. In this case, a measure of theflocculation or coalescence in an emulsion can be obtained using a simple UV-visiblespectrophotometer (Walstra 1968, Pearce and Kinsella 1978, Reddy and Fogler 1981, Pandolfeand Masucci 1984). The turbidity of light is measured at a single wavelength or over a rangeof wavelengths, and the mean particle size is estimated using light-scattering theory. Thistechnique should be used with caution, because the relationship between particle size andturbidity is fairly complex in the region where the droplet radius is the same order ofmagnitude as the wavelength of light used (see Figure 9.12).

10.2.3. Interfacial Tension

One of the most valuable means of obtaining information about the characteristics of anemulsifier is to measure the reduction in the surface (or interfacial) tension when it adsorbsto a surface (or an interface). Surface tension measurements can be used to determine thekinetics of emulsifier adsorption, the packing of emulsifier molecules at an interface, criticalmicelle concentrations, surface pressures, and competitive adsorption (Chapter 5). A varietyof analytical instruments are available for measuring the surface or interfacial tension ofliquids, as discussed in Section 5.10.

10.2.4. Interfacial Rheology

The stability of food emulsions to creaming and droplet coalescence depends on the rheologi-cal characteristics of the interfacial membranes which surround the droplets (Chapter 7), andso it is often important for food scientists to be able to quantify the rheological characteristicsof interfaces. In addition, interfacial rheology measurements can also be used to providevaluable information about other characteristics of emulsifiers, such as adsorption kinetics,competitive adsorption, and interfacial interactions. Interfacial rheology is the two-dimen-sional equivalent of bulk rheology (Chapter 8), and consequently many of the principles andconcepts are directly analogous.* An interface can be viscous, elastic, or viscoelastic depend-ing on the type, concentration, and interactions of the molecules present (Murray and Dickinson

* Although it is convenient to consider interfacial rheology in two dimensions, it must be remembered that theinterfacial membrane has a finite thickness (usually a few nanometers) in reality, and this may also influence itsrheological characteristics.


1996). Two types of deformation are particularly important at an interface: shear and dilata-tion. The shear behavior of an interface is characterized by its resistance to the “sliding” ofneighboring regions past one another, without any change in the overall interfacial area. Thedilatational behavior of an interface is characterized by its resistance to the expansion orcontraction of its surface area. Instruments for measuring the shear and dilatational rheologyof interfaces were reviewed in Section 5.11.

10.3. MICROSTRUCTURE AND DROPLET SIZE DISTRIBUTION

10.3.1. Microscopy

The unaided human eye can resolve objects which are greater than about 0.1 mm (100 µm)apart (Aguilera and Stanley 1990). Many of the structural components in food emulsions aresmaller than this lower limit and therefore cannot be observed directly by the eye (e.g.,emulsion droplets, surfactant micelles, fat crystals, gas bubbles, and protein aggregates)(Dickinson 1992). Our normal senses must therefore be augmented by microscopic tech-niques which enable us to observe tiny objects (Aguilera and Stanley 1990, Kalab et al. 1995,Smart et al. 1995). A number of these techniques are available to provide information aboutthe structure, dimensions, and organization of the components in food emulsions (e.g., opticalmicroscopy, scanning and transmission electron microscopy, and atomic force microscopy)(Kirby et al. 1995, Kalab et al. 1995, Smart et al. 1995). These techniques have the abilityto provide information about structurally complex systems in the form of “images” which arerelatively easy to comprehend by human beings (Kirby et al. 1995). Each microscopictechnique works on different physicochemical principles and can be used to examine differ-ent levels and types of structural organization. Nevertheless, any type of microscope musthave three qualities if it is going to be used to examine the structure of small objects:resolution, magnification, and contrast (Aguilera and Stanley 1990). Resolution is the abilityto distinguish between two objects which are close together. Magnification is the number oftimes that the image is greater than the object being examined. Contrast determines how wellan object can be distinguished from its background.

10.3.1.1. Conventional Optical Microscopy

Although the optical microscope was developed over a century ago, it is still one of the mostvaluable tools for observing the microstructure of emulsions (Mikula 1992; Hunter 1986,1993). An optical microscope contains a series of lenses which direct the light through thespecimen and magnify the resulting image (Figure 10.1). The resolution of an optical micro-scope is determined by the wavelength of light used and the mechanical design of theinstrument (Franklin 1977, Hunter 1986). The theoretical limit of resolution of an opticalmicroscope is about 0.2 µm, but in practice it is difficult to obtain reliable measurementsbelow about 1 µm (Hunter 1993). This is because of technical difficulties associated with thedesign of the instrument and because the Brownian motion of small particles causes imagesto appear blurred. The optical microscope therefore has limited application to many foodemulsions because they contain structures with sizes below the lower limit of resolution.Nevertheless, it can provide valuable information about the size distribution of droplets inemulsions which contain larger droplets and can be used to distinguish between flocculationand coalescence (Mikula 1992), which is often difficult using instrumental techniques basedon light scattering, electrical pulse counting, or ultrasonics.

The natural contrast between the major components in food emulsions is often fairly poor(because they have similar refractive indices or color), which makes it difficult to reliablydistinguish them from each other using conventional bright-field optical microscopy. For this


FIGURE 10.1 Comparison of conventional bright-field and dark-field microscopes.

reason, the technique has been modified in a number of ways to enhance the contrast, improvethe image quality, and provide more detailed information about the composition and micro-structure of food emulsions. Various types of chemical stains are available which bind toparticular components within an emulsion (e.g., the proteins, polysaccharides, or lipids) andtherefore enable specific structural features to be highlighted (Gurr 1961, Smart et al. 1995).These stains must be used with caution because they can alter the structure being examined.In addition, they are often difficult to incorporate into concentrated or semisolid emulsions.

The contrast between different components can be improved without using chemical stainsby modifying the design of the optical microscope (e.g., by using phase contrast or differ-ential interference contrast microscopy) (Aguilera and Stanley 1990). These techniquesimprove contrast by using special lenses which convert small differences in refractive indexinto differences in light intensity.

The structure of optically anisotropic food components, such as fat crystals, starch gran-ules, and muscle fibers, can be studied using birefringent microscopy (Aguilera and Stanley1990). Anisotropic materials cause plane polarized light to be rotated, whereas isotropicmaterials do not, and so it is possible to distinguish anisotropic and isotropic materials using“crossed polarizers.” This technique is particularly useful for monitoring phase transitions offat and for determining the location and morphology of fat crystals in emulsion droplets(Walstra 1967, Boode 1992).

The characteristics of particles with sizes less than a micrometer can be observed by dark-field illumination using an instrument known as an ultramicroscope (Shaw 1980, Farinato andRowell 1983, Hunter 1986). A beam of light is passed through the specimen at a right angleto the eyepiece (Figure 10.1). In the absence of any particles, the specimen appears com-pletely black, but when there are particles present, they scatter light and the image appearsas a series of bright spots against a black background. This technique can be used to detectparticles as small as 10 nm; however, the particles appear as blurred spots rather than well-defined images whose size can be measured directly. The particle size is inferred from thebrightness of the spots or from measurements of their Brownian motion. The ultramicroscopecan also be used to determine the number of particles in a given volume or to monitor themotion of particles in an electric field (Section 10.6).

Certain food components either fluoresce naturally or can be made to fluoresce by addingfluorescent dyes which bind to them (Aguilera and Stanley 1990, Kalab et al. 1995).


Fluorescent materials adsorb electromagnetic radiation at one wavelength and emit it at ahigher wavelength (Skoog et al. 1994). A conventional bright-field optical microscope canbe modified to act as a fluorescence microscope by adding two filters (or other suitablewavelength selectors). One filter is placed before the light enters the sample and producesa monochromatic excitation beam. The other filter is placed after the light beam has passedthrough the sample and produces a monochromatic emission beam. A variety of fluorescentdyes (fluorophores) are available which bind to specific components within a food (e.g.,proteins, fats, or carbohydrates) (Larison 1992). Fluorescence microscopes usually use anultraviolet light source to illuminate the specimen (which is therefore invisible to the humaneye), whereas the light emitted by the fluorescent components within a specimen is in thevisible part of the electromagnetic spectrum, and so they appear as bright objects against ablack background. Fluorescence microscopy is a very sensitive technique that is particularlyuseful for studying structures that are present at such small concentrations that they cannotbe observed using conventional optical microscopy. In addition, it can be used to highlightspecific structures within an emulsion by selecting fluorescent dyes which bind to them.

One of the major drawbacks of optical microscopy is the possibility that sample prepa-ration alters the structure of the specimen being analyzed (Aguilera and Stanley 1990, Kalabet al. 1995). Sample preparation may be a simple procedure, such as spreading an emulsionacross a slide, or a more complex procedure, such as fixing, embedding, slicing, and staininga sample (Smart et al. 1995). Even a procedure as simple as spreading a specimen acrossa slide may alter its structural properties and should therefore be carried out carefully andreproducibly. Other disadvantages of optical microscopy are that measurements are oftentime consuming and subjective, it is often necessary to analyze a large number of differentregions within a sample to obtain statistically reliable data, and it is limited to studyingstructures greater than about 1 µm (Mikula 1992). Many modern optical microscopes nowhave the capability of being linked to personal computers which can rapidly store andanalyze images and thus enhance their ease of operation (Klemaszeski et al. 1989, Mikula1992).

10.3.1.2. Laser Scanning Confocal Microscopy

This is a fairly recent development in optical microscopy which can provide extremelyvaluable information about the microstructure of food emulsions (Blonk and van Aalst 1993,Brooker 1995, Smart et al. 1995, Vodovotz et al. 1996). Laser scanning confocal microscopy(LSCM) provides higher clarity images than conventional optical microscopy and allows thegeneration of three-dimensional images of structures without the need to physically sectionthe specimen. The LSCM focuses an extremely narrow laser beam at a particular point in thespecimen being analyzed, and a detector measures the intensity of the resulting signal (Figure10.2). A two-dimensional image is obtained by carrying out measurements at different pointsin the x–y plane, either by moving the specimen (and keeping the laser beam stationary) orby moving the laser beam (and keeping the specimen stationary). An image is generated bycombining the measurements from each individual point. Three-dimensional images areobtained by focusing the laser beam at different depths into the sample and then scanning inthe horizontal plane. Observation of the microstructure of multicomponent systems is oftenfacilitated by using the natural fluorescence of certain components or by using fluorescentdyes that bind selectively to specific components (e.g., proteins, lipids, or polysaccharides)(Larison 1992). The LSCM technique suffers from many of the same problems as conven-tional optical microscopy. Nevertheless, it has a slightly better resolution and sensitivity, andthe sample preparation is often less severe.

LSCM has been used to study the size, concentration, and organization of droplets inemulsions (Jokela et al. 1990); to examine the microstructure of butter, margarines, and low-


FIGURE 10.2 Laser scanning confocal microscopy. A laser beam is scanned across a particular x–yplane within a sample. By combining successive x–y planes, it is possible to create a three-dimensionalimage.

fat spreads (Blonk and van Aalst 1993); to monitor creaming of droplets (Brakenoff et al.1988); and to follow the desorption of proteins from the surface of emulsion droplets.

10.3.1.3. Electron Microscopy

Electron microscopy is widely used to examine the microstructure of food emulsions, particu-larly those that contain structural components which are smaller than the lower limit ofresolution of optical microscopes (ca. <1 µm) (e.g., protein aggregates, small emulsiondroplets, fat or ice crystals, micelles, and interfacial membranes) (Chang et al. 1972, Tungand Jones 1981, Hunter 1986, Aguilera and Stanley 1990, Heertje and Paques 1995). It canbe used to provide information about the concentration, dimensions, and spatial distributionof structural entities within a specimen, provided the microstructure of the specimen is notsignificantly altered by the sample preparation (Heertje and Paques 1995). With suitablesample preparation, electron microscopy can be used to analyze both oil-in-water and water-in-oil emulsions that are either liquid or solid. Electron microscopes are fairly large piecesof equipment which are relatively expensive to purchase and maintain (Smart et al. 1995). Forthis reason, they tend to be available only at fairly large research laboratories or foodcompanies.

Electron microscopes use electron beams, rather than light beams, to provide informationabout the structure of materials (Aguilera and Stanley 1990, Heertje and Paques 1995). Thesebeams are directed through the microscope using a series of magnetic fields rather thanoptical lenses. Electron beams have much smaller wavelengths than light and so can be usedto examine much smaller objects. In principle, the smallest size that can be resolved is about0.2 nm, but in practice, it is usually about 1 nm due to limitations in the stability andperformance of the magnetic lenses. Two types of electron microscope are commonly usedto examine the structure of food systems: transmission electron microscopy (TEM) andscanning electron microscopy (SEM). In both of these techniques, it is necessary to keep themicroscope under high vacuum because electrons are easily scattered by atoms or moleculesin a gas, and this would cause a deterioration in the image quality. This also means thatspecimens must be prepared so that they are free of all volatile components that couldevaporate (e.g., water and organic molecules).

Transmission Electron Microscopy. A cloud of electrons, produced by a tungsten cathode,is accelerated through a small aperture in a positively charged plate to form an electron beam(Figure 10.3). This beam is focused and directed through the specimen by a series ofmagnetic lenses. Part of the electron beam is either adsorbed or scattered by the specimen,while the rest is transmitted. The beam of transmitted electrons is magnified by a magnetic


FIGURE 10.3 Schematic diagram of TEM. An electron beam is directed through the sample, and theintensity of the transmitted beam is measured.

lens and then projected onto a fluorescent screen to create an image of the specimen. Thefraction of electrons transmitted by a substance depends on its electron density: the lowerthe electron density, the greater the fraction of electrons transmitted and the more intensethe image. Components with different electron densities therefore appear as regions ofdifferent intensity on the image. Images are typically 100 to 500,000 times larger than theportion of specimen being examined, which means that structures as small as 0.4 nm can beobserved (Hunter 1993).

Electrons are highly attenuated by most materials, and therefore specimens must be ex-tremely thin in order to allow enough of the electron beam through to be detected (Aguileraand Stanley 1990, Heertje and Paques 1995). Specimens used in TEM are therefore muchthinner (≈0.05 to 0.1 µm) than those used in light microscopy (≈1 µm to a few millimeters).The difference in electron densities of food components is quite small, and therefore it isdifficult to distinguish between them. For this reason, the density contrast between compo-nents is usually enhanced by selectively staining the sample with various heavy metal salts(which have high electron densities), such as lead, tungsten, or uranium. These salts bindselectively to specific components, which enables them to be distinguished from the rest ofthe sample. The metal salts may bind to the component itself (positive staining) or to thesurrounding material (negative staining). In positive staining, a dark specimen is seen againsta light background, whereas in negative staining, an illuminated specimen is seen against adark background. The need to have very thin and dehydrated specimens, that often requirestaining, means that sample preparation is considerably more time consuming, destructive,and cumbersome than for other forms of microscopy.

The information produced by TEM is usually in the form of a two-dimensional image thatrepresents a thin slice of the specimen (Figure 10.4). Nevertheless, it is possible to obtainsome insight about the three-dimensional structure of a sample using a technique called metalshadowing (Figure 10.5). A vapor of a heavy metal, such as platinum, is sprayed onto thesurface of a sample at an angle (Hunter 1993). The sample is then dissolved away using astrong acid, which leaves a metal replica of the sample. When a beam of electrons is


(a)

FIGURE 10.4 Comparison of electron micrographs of emulsions produced by (a) SEM (of butter) and(b) TEM (of margarine). TEM normally produces a two-dimensional image of a sample, whereas SEMproduces a more three-dimensional image. The SEM of butter: g = fat droplet, f = fat crystals. The TEMof margarine: W = water droplets, F = fat continuous matrix, c = fat crystals. (Photographs kindlyprovided by I. Heertje.)

(b)


FIGURE 10.5 TEM can be used to obtain images of the surface topography of a sample using atechnique known as metal shadowing.

transmitted through the metal replica, the “shadows” formed by the specimen are observedas illuminated regions which have characteristic patterns from which the topography of thespecimen can be deduced.

Scanning Electron Microscopy. SEM is used to provide images of the surface topographyof specimens (Heertje and Paques 1995). It relies on the measurement of secondary electronsgenerated by a specimen when it is bombarded by an electron beam, rather than the electronswhich have traveled through the specimen. A focused electron beam is directed at a particularpoint on the surface of a specimen. Some of the energy associated with the electron beam isabsorbed by the material and causes it to generate secondary electrons, which leave thesurface of the sample and are recorded by a detector. An image of the specimen is obtainedby scanning the electron beam in an x–y direction over its surface and recording the numberof electrons generated at each location. Because the intensity at each position depends on theangle between the electron beam and the surface, the electron micrograph has a three-dimensional appearance (Figure 10.4).

Sample preparation for SEM is considerably easier and tends to produce fewer artifactsthan TEM. Because an image is produced by secondary electrons generated at the surface ofa specimen, rather than by an electron beam that travels through a specimen, it is notnecessary to use ultrathin samples. Even so, specimens often have to be cut, fractured, fixed,and dehydrated, which may alter their structures. The resolving power of SEM is about 3 to4 nm, which is an order of magnitude worse than TEM but about three orders of magnitudebetter than optical microscopy. Another major advantage of SEM over optical microscopy isthe large depth of field, which means that images of relatively large structures are all in focus(Hunter 1993).

Electron microscopy is widely used by food scientists to determine the size of emulsiondroplets; the dimensions and structure of flocs; the surface morphology of droplets and airbubbles; the size, shape, and location of fat crystals; and the microstructure of three-dimen-sional networks of aggregated biopolymers (Chang et al. 1972, Kalab 1981, Tung and Jones


FIGURE 10.6 Principles of AFM. A probe is scanned over the surface of a material, and its displace-ment is measured using an accurate laser detection system. The force required to keep the probe in thesame location, or the deflection of the probe, is measured.

1981, Hermansson 1988, Aguilera and Stanley 1990, Bucheim and Dejmek 1990, Lee andMorr 1992, Heertje and Paques 1995). TEM and SEM electron micrographs of two water-in-oil emulsions are shown in Figure 10.4. TEM gives a two-dimensional cross-section of thesample, whereas SEM gives a more three-dimensional image. As mentioned earlier, the majorlimitation of the technique is the difficulty in preparing samples without altering theirstructure. In addition, the use of high-energy electron beams can change the structure ofdelicate specimens. Many of these problems are being overcome as the result of recentadvances in the design of electron microscopes and sample preparation techniques (Smart etal. 1995).

10.3.1.4. Atomic Force Microscopy

Atomic force microscopy (AFM) has the ability to provide information about structures at theatomic and molecular levels and is therefore complementary to the other forms of microscopymentioned above (Miles and McMaster 1995). The technique has only recently been devel-oped as a commercial instrument, although these instruments are still fairly expensive topurchase. For this reason, the application of AFM to foods is still largely in its infancy (Kirbyet al. 1995) and is only used by a small number of research laboratories. Nevertheless, thetechnique has the potential to provide a wealth of valuable information about the structure andorganization of molecules in food emulsions and will increase our fundamental understandingof these complex systems.

The AFM creates an image by scanning a tiny probe (similar to the stylus of a recordplayer, but only a few micrometers in size) across the surface of the specimen beinganalyzed (Figure 10.6). When the probe is held extremely close to the surface of a material,it experiences a repulsive force, which causes the cantilever to which it is attached to be bentaway from the surface. The extent of the bending is measured using an extremely sensitiveoptical system. By measuring the deflection of the stylus as it is moved over the surface ofthe material, it is possible to obtain an image of its structure. In practice, it is more commonto measure the force required to keep the deflection of the stylus constant, as this reducesthe possible damage caused by a stylus as it moves across the surface of a sample. Theresolution of AFM depends principally on the size and shape of the probe and the accuracyto which it can be positioned relative to the sample. Samples to be analyzed are usuallydissolved in a suitable solvent and then dried onto the surface of an extremely flat plate suchas mica.

Atomic force measurements have been used to observe the structure of individual andaggregated polysaccharide and protein molecules (e.g., xanthan, pectin, acetan, starch, col-


lagen, myosin, and bovine serum albumin) (Miles and McMaster 1995, Kirby et al. 1995).This type of study is useful for examining the relationship between the structure and inter-actions of biopolymer molecules and the type of gels they form. AFM has also been used tostudy the molecular organization of emulsifiers at planar oil–water and air–water interfaces(Kirby et al. 1995).

10.3.2. Static Light Scattering

10.3.2.1. Principles

Static light scattering is used to determine particle sizes between about 0.1 and 1000 µm andis therefore suitable for characterizing the droplets in most food emulsions. When a beamof light is directed through an emulsion, it is scattered by the droplets (Dickinson andStainsby 1982, Farinato and Rowell 1983, Hiemenz 1986, Hunter 1986, Everett 1988). Ameasurement of the degree of scattering can be used to provide information about the dropletsize distribution and concentration. Analytical instruments based on this principle have beencommercially available for many years (Mikula 1992) and are widely used in the foodindustry for research, development, and quality control purposes. Most of these instrumentsare fully automated, simple to use, and provide an analysis of an emulsion within a fewminutes. Even so, they tend to be fairly expensive to purchase, which has limited theirapplication somewhat.

The interaction of an electromagnetic wave with an emulsion is characterized by a scat-tering pattern, which is the angular dependence of the intensity of the light emerging fromthe emulsion, I (Φ) (Figure 10.7). The size and concentration of droplets in an emulsion areascertained from the scattering pattern using a suitable theory (Farinato and Rowell 1983).Theories that relate light-scattering data to droplet size distributions are based on a math-ematical analysis of the propagation of an electromagnetic wave through an ensemble ofparticles (van de Hulst 1957, Bohren and Huffman 1983). A number of theories are available,which vary according to their mathematical complexity and the type of systems to which theycan be applied. The interaction between light waves and emulsion droplets can be conve-niently divided into three regimes, according to the relationship between the droplet radius(r) and the wavelength (λ): (1) long-wavelength regime (r < λ/20), (2) intermediate-wave-length regime (λ/20 < r < 20λ), and (3) short-wavelength regime (r > 20λ). A characteristicscattering pattern is associated with each of these regimes (Figure 10.8).

FIGURE 10.7 The scattering pattern from an emulsion is characterized by measuring the intensity ofthe scattered light, I (Φ), as a function of angle (Φ) between the incident and scattered beams. Thescattering pattern is particularly sensitive to particle size relative to the wavelength of light.


FIGURE 10.8 Typical scattering patterns for droplets in the long-, intermediate-, and short-wave-length regimes.

Long-Wavelength Regime. The simplest equation for relating the characteristics of the par-ticles in a suspension to the scattering pattern produced when a monochromatic light beampasses through it was derived by Lord Rayleigh over a century ago and is applicable in thelong-wavelength regime (Sherman 1968b):

I

I

r

R

n n

n ni

( )( cos )

ΦΦ=

−( )+( )

+6

21

3 3

4 2

12

02 2

12

02 2

2π φλ (10.1)

where Ii is the initial intensity of the light beam in the surrounding medium, φ is the dispersed-phase volume fraction, R is the distance between the detector and the scattering droplet, andn0 and n1 are the refractive indices of the continuous phase and droplets, respectively. Thisequation cannot be used to interpret the scattering patterns of most food emulsions becausethe size of the droplets (typically between 0.1 and 100 µm) is of the same order or larger thanthe wavelength of light used (typically between 0.2 and 1 µm). Nevertheless, it can be appliedto suspensions which contain smaller particles, such as surfactant micelles or protein mol-ecules (Hiemenz 1986). Equation 10.1 also provides some useful insights into the factorswhich influence the scattering profile of emulsions. It indicates that the degree of scatteringfrom an emulsion is linearly related to the droplet concentration and increases as the refrac-tive indices of the materials become more dissimilar.

Intermediate-Wavelength Regime. As mentioned earlier, most food emulsions contain drop-lets which are in the intermediate-wavelength regime. The scattering profile in this regime isextremely complex because light waves scattered from different parts of the same droplet areout of phase and therefore constructively and destructively interfere with one another (Shaw1980, Hiemenz 1986). For the same reason, the mathematical relationship between thescattering pattern and the particle size is much more complex. A mathematician called Miedeveloped a theory which can be used to interpret the scattering patterns of dilute emulsionsthat contain spherical droplets of any size (van de Hulst 1957, Kerker 1969). The Mie theoryis fairly complicated, but it can be solved rapidly using modern computers. This theory givesexcellent agreement with experimental measurements and is used by most commercial par-ticle-sizing instruments. It should be pointed out that the Mie theory assumes that the lightwaves are only scattered by a single particle, and so it is only strictly applicable to diluteemulsions. In more concentrated emulsions, a light beam scattered by one droplet maysubsequently interact with another droplet, and this alters the scattering pattern (Ma et al.1990). For this reason, emulsions must be diluted prior to analysis to a concentration wheremultiple scattering effects are negligible (i.e., φ < 0.05%).

Short-Wavelength Regime. When the wavelength of light is much smaller than the particlediameter, the droplet size distribution can be determined directly by optical microscopy(Section 10.3.1).


FIGURE 10.9 Design for a particle-sizing instrument that utilizes angular scattering.

10.3.2.2. Measurement Techniques

Angular Scattering Methods. Most modern particle-sizing instruments, based on staticlight scattering, measure the angular dependence of the scattered light (Farinato and Rowell1983, Mikula 1992). The sample to be analyzed is diluted to an appropriate concentrationand then placed in a glass measurement cell (Figure 10.9). A laser beam is generated by ahelium–neon laser (λ = 632.8 nm) and directed through the measurement cell, where it isscattered by the emulsion droplets. The intensity of the scattered light is measured as afunction of scattering angle using an array of detectors located around the sample. Thescattering pattern recorded by the detectors is sent to a computer, where it is stored andanalyzed. The droplet size distribution that gives the best fit between the experimentalmeasurements of I (Φ) versus Φ and those predicted by the Mie theory is calculated. Thedata are then presented as a table or graph of droplet concentration versus droplet size(Section 1.3.2). Once the emulsion sample has been placed in the instrument, the measure-ment procedure is fully automated and only takes a few minutes to complete. Beforeanalyzing a sample, the instrument is usually blanked by measuring the scattering profilefrom the continuous phase in the absence of emulsion droplets. This scattering pattern isthen subtracted from that of the emulsion to eliminate extraneous scattering from back-ground sources other than the droplets (e.g., dust or optical imperfections). The range ofdroplet radii that can be detected using this type of experimental arrangement is about 0.1to 1000 µm. A number of instrument manufacturers have developed methods of determiningdroplet sizes down to 0.01 µm, but there is still much skepticism about the validity of dataat these low sizes. This is highlighted by the fact that measurements on exactly the sameemulsion using a number of different commercial instruments often give very differentparticle size distributions (Coupland and McClements 1998).

Spectroturbidimetric Methods. These techniques measure the turbidity of a dilute emulsionas a function of wavelength (Walstra 1968, Pearce and Kinsella 1978, Reddy and Fogler1981, Pandolfe and Masucci 1984). The turbidity (τ) is determined by comparing the intensityof light which has traveled directly through an emulsion (I ) with that which has traveleddirectly through the pure continuous phase (Ii): τ = –ln(I/Ii)/d, where d is the sample pathlength. The greater the scattering of light by an emulsion, the lower the intensity of thetransmitted wave, and therefore the larger the turbidity. The turbidity of a dilute emulsion islinearly related to the dispersed-phase volume fraction, and so turbidity measurements can be


used to determine φ if the droplet size remains constant. Turbidity versus wavelength spectrafor emulsions with different droplet sizes are shown in Figure 10.10.

The emulsion to be analyzed is placed in a cuvette, and its turbidity is measured over arange of wavelengths (typically between 200 to 1000 nm). The droplet size distribution isthen determined by finding the best fit between the experimental measurements of turbidityversus wavelength and those predicted by the Mie theory. The spectroturbidimetric techniquecan be carried out using the UV-visible spectrophotometers found in most research labora-tories, which may circumnavigate the need to purchase one of the expensive commerciallight-scattering instruments mentioned above. Nevertheless, some samples adsorb light stronglyin the UV-visible region, which interferes with the interpretation of the turbidity spectra. Inaddition, the refractive indices of the dispersed and continuous phases must be known, andthese vary with wavelength (Walstra 1968).

Reflectance Methods. An alternative method of determining the droplet size of emulsionsis to measure the light reflected back from the emulsion droplets (Φ = 180°). The intensityof backscattered light is related to the size of the droplets in an emulsion (Lloyd 1959,Sherman 1968b). This technique has been used much less frequently than thespectroturbidimetric or angular scattering techniques but may prove useful for the study ofmore concentrated emulsions which are opaque to light.

10.3.2.3. Applications

The principal application of static light-scattering techniques in the food industry is todetermine droplet size distributions (Dickinson and Stainsby 1982). A knowledge of theparticle size of an emulsion is useful for predicting its long-term stability to creaming,flocculation, coalescence, and Ostwald ripening (Chapter 7). Measurements of the timedependence of the particle size distribution can be used to monitor the kinetics of theseprocesses (Chapter 7). Light-scattering instruments are widely used in research and develop-ment laboratories to investigate the influence of droplet size on physicochemical properties,such as stability, appearance, and rheology. They are also used in quality control laboratoriesto ensure that a product meets the relevant specifications for droplet size.

It should be noted that the data from any commercial light-scattering technique should betreated with some caution. To determine the droplet size distribution of an emulsion, com-

FIGURE 10.10 Turbidity versus wavelength spectra for emulsions with different droplet sizes.


mercial instruments have to make some a priori assumption about the shape of the distribu-tion in order to solve the scattering theory in a reasonable time. In addition, the solution ofthe scattering theory is often particularly sensitive to the refractive indices and adsorptivitiesof the continuous and dispersed phases, and these values are often not known accurately(Zhang and Xu 1992). Finally, the mechanical design of each commercial instrument isdifferent. All of these factors mean that the same emulsion can be analyzed using instrumentsfrom different manufacturers (or even from the same manufacturer) and quite large variationsin the measured droplet size distributions can be observed, even though they should beidentical (Coupland and McClements 1998). For this reason, commercial light-scatteringinstruments are often more useful for following qualitative changes rather than giving abso-lute values.

One must be especially careful when using light scattering to determine the particle sizedistribution of flocculated emulsions. The theory used to calculate the size distributionassumes that the particles are isolated homogeneous spheres. In flocculated emulsions, thedroplets aggregate into heterogeneous “particles” which have an ill-defined refractive indexand shape. Consequently, the particle size distribution determined by light scattering givesonly an approximate indication of the true size of the flocs. In addition, emulsions often haveto be diluted in continuous phase (to eliminate multiple scattering effects) and stirred (toensure they are homogeneous) prior to measurement. Dilution and stirring are likely todisrupt any weakly flocculated droplets but leave strongly flocculated droplets intact. Forthese reasons, commercial light-scattering techniques can only give a qualitative indicationof the extent of droplet flocculation. Another important limitation of light-scattering tech-niques is that they cannot be used to analyze optically opaque or semisolid emulsions in situ(e.g., butter, margarine, and ice cream).

10.3.3. Dynamic Light Scattering


Dynamic light scattering is used to determine the size of particles which are below the lowerlimit of detection of static light-scattering techniques (e.g., small emulsion droplets, proteinaggregates, and surfactant micelles) (Hallet 1994, Dalgleish and Hallet 1995, Horne 1995).Instruments based on this principle are commercially available and are capable of analyz-ing particles with diameters between about 3 nm and 3 µm. Dynamic light-scattering tech-niques utilize the fact that the droplets in an emulsion continually move around because oftheir Brownian motion (Hunter 1986). The translational diffusion coefficient (D) of thedroplets is determined by analyzing the interaction between a laser beam and the emulsion,and the size of the droplets is then calculated using the Stokes–Einstein equation (Horne1995):

rkT

D=

6 1πη (10.2)

where η1 is the viscosity of the continuous phase.


A number of experimental techniques have been developed to measure the translationaldiffusion coefficient of colloidal particles (Hunter 1993). The two most commonly usedmethods in commercial instruments are photon correlation spectroscopy (PCS) and Dopplershift spectroscopy (DSS).


Photon Correlation Spectroscopy. When a laser beam is directed through an ensemble ofparticles, a scattering pattern is produced which is a result of the interaction between theelectromagnetic waves and the particles (Horne 1995). The precise nature of this scatteringpattern depends on the relative position of the particles in the measurement cell. If thescattering pattern is observed over very short time intervals (approximately microseconds),one notices that there are slight variations in its intensity with time, which are caused by thechange in the relative position of the particles due to their Brownian motion. The frequencyof these fluctuations depends on the speed at which the particles move, and hence on theirsize. The change in the scattering pattern is monitored using a detector that measures theintensity of the photons, I (t), which arrive at a particular scattering angle (or range ofscattering angles) with time. If there is little change in the position of the particles within aspecified time interval (τ), the scattering pattern remains fairly constant and I (t) ≈ I (t + τ).On the other hand, if the particles move an appreciable distance within the time interval, thescattering pattern is altered significantly and I (t) ≠ I (t + τ). The correlation between thescattering patterns can be expressed mathematically by an autocorrelation function (Horne1995):

CN

I t I ti ii

N

( ) ( ) ( )τ τ= +=∑1

1(10.3)

where N is the number of times this procedure is carried out, which is typically of the orderof 105 to 106. As the time interval (τ) between which the two scattering patterns arecompared is increased, the autocorrelation function decreases from a high value, where thescattering patterns are highly correlated, to a constant low value, when all correlationbetween the scattering patterns is lost (Figure 10.11). For a monodisperse emulsion, theautocorrelation function decays exponentially with a relaxation time (tc) that is related to thetranslational diffusion coefficient of the particles: tc = (2Q2D)–1, where Q is the scatteringvector, which describes the strength of the interaction between the light wave and theparticles (Horne 1995):

Qn= 4

2πλ

θsin (10.4)

FIGURE 10.11 Decay of the autocorrelation function with time interval due to Brownian motion ofthe particles. The decay occurs more rapidly for small particles because they move faster.


FIGURE 10.12 The distribution of frequency shifts can be related to the particle size distributionusing suitable theories: small particles move more rapidly and therefore cause a greater frequency shift.

Here, n is the refractive index of the medium, λ is the wavelength of light, and θ is thescattering angle. By measuring the decay of the autocorrelation function with time, it ispossible to determine the diffusion coefficient and hence the particle size from the Stokes–Einstein equation (Equation 10.2).

PCS is suitable for accurate determination of particle size in suspensions that are mono-disperse or that have a narrow size distribution. It is less reliable for suspensions with broadsize distributions because of difficulties associated with interpreting the more complexautocorrelation decay curves (Horne 1995). PCS can be used to monitor the flocculation ofparticles in suspensions because aggregation causes them to move more slowly (Dalgleishand Hallet 1995). It can also be used to determine the thickness of adsorbed layers ofemulsifier on spherical particles (Dalgleish and Hallet 1995). The radius of the sphericalparticles is measured in the absence of emulsifier (provided they are stable to aggregation)and then in the presence of emulsifier. The difference in radius is equal to the thickness ofthe adsorbed layer, although this value may also include the presence of any solvent mol-ecules associated with the emulsifier. PCS is restricted to the analysis of dilute suspensionsof particles (φ < 0.1%).

Doppler Shift Spectroscopy. When a laser beam is scattered from a moving particle, itexperiences a shift in frequency, known as a Doppler shift (Trainer et al. 1992, Horne 1995).The frequency of the scattered wave increases slightly when the particle moves toward thelaser beam and decreases slightly when it moves away. Consequently, there is a symmetricaldistribution of Doppler shifts around the original frequency of the laser beam. The magnitudeof the Doppler shift increases with particle velocity, and hence with decreasing particle size.Consequently, there is a broad distribution of Doppler shifts in a polydisperse suspension thatcontains particles moving at different velocities (Figure 10.12). The particle size distributionis determined by analyzing this Doppler shift spectrum. Two types of measurement tech-niques are commonly used in DSS: homodyne and heterodyne. Homodyne techniques deter-mine the Doppler shifts by making use of the interference of light scattered from one particlewith that scattered from all the other particles, whereas heterodyne techniques determinefrequency shifts by comparing the frequency of the light scattered from the particles with areference beam of fixed frequency (Trainer et al. 1992). The heterodyne technique is capableof analyzing suspensions with much higher particle concentrations than is possible using thehomodyne technique, and so only it will be considered here.

A typical experimental arrangement for making Doppler shift measurements is shown inFigure 10.13. A laser beam is propagated along an optical waveguide which is immersed inthe sample being analyzed. Part of the laser beam is reflected from the end of the waveguideand returns to the detector, where it is used as a reference beam because its frequency is not


FIGURE 10.13 Experimental technique for measuring the Doppler shift of particles based on theheterodyne measurement technique.

Doppler shifted. The remainder of the laser beam propagates into the sample, and part of itis scattered back up the waveguide by the particles in its immediate vicinity (≈100 µm pathlength). This backscattered light is frequency shifted by an amount that depends on thevelocity of the particles. The difference in frequency between the reflected and scatteredwaves is equivalent to the Doppler shift.

The variation of the power, P(ω), of the scattered waves with angular frequency (ω) hasthe form of a Lorentzian function:

P I IS( )ωω

ω ω=

+00

202 (10.5)

where I0 and <IS> are the reference and scattered light intensities, and ω0 is a characteristicfrequency that is related to the scattering efficiency and diffusion coefficient of the particles:ω0 = DQ2. Particle size is determined by measuring the variation of P(ω) with ω, and thenfinding the value of the diffusion coefficient which gives the best fit between the measuredspectra and that predicted by Equation 10.5. For polydisperse systems, it is necessary to takeinto account that there is a distribution of particle sizes. The calculation of particle sizerequires that the analyst input the refractive indices of the continuous and dispersed phasesand the viscosity of the continuous phase. Because the path length of the light beam in thesample is so small (about 100 µm), it is possible to analyze much more concentratedemulsions (up to 40%) than with static light scattering or PCS techniques (Trainer et al.1992), although some form of correction factor usually has to be applied to take into accountthe influence of the droplet interactions on the diffusion coefficient (Horne 1995). Commer-cial instruments are available that are easy to use and which can analyze an emulsion in a fewminutes. The major limitations of dynamic light-scattering techniques are that they can onlybe used to analyze droplets with sizes smaller than about 3 µm and that the viscosity of theaqueous phase must be Newtonian, which is not the case for many food emulsions, especiallythose that contain thickening agents.

10.3.4. Electrical Pulse Counting

Electrical pulse counting techniques are capable of determining the size distribution ofparticles with diameters between about 0.6 and 400 µm (Hunter 1986, Mikula 1992) and aretherefore suitable for analyzing most food emulsions. These instruments have been commer-cially available for many years and are widely used in the food industry. The emulsion to beanalyzed is placed in a beaker that has two electrodes dipping into it (Figure 10.14). One of


FIGURE 10.14 An electrical pulse counter. The decrease in current is measured when an oil dropletpasses through a small hole in a glass tube.

the electrodes is contained in a glass tube which has a small hole in it, through which theemulsion is sucked. When an oil droplet passes through the hole, it causes a decrease in thecurrent between the electrodes because oil has a much lower electrical conductivity thanwater. Each time a droplet passes through the hole, the instrument records a decrease incurrent, which it converts into an electrical pulse. The instrument controls the volume ofliquid that passes through the hole, and so the droplet concentration can be determined bycounting the number of electrical pulses in a known volume. When the droplets are smallcompared to the diameter of the hole, the droplet size is simply related to the height of thepulses: d3 = kP, where d is the droplet diameter, P is the pulse height, and k is an instrumentconstant which is determined by recording the pulse height of a suspension of monodisperseparticles of known diameter.

To cover the whole range of droplet sizes from 0.6 to 400 µm, it is necessary to use glasstubes with different sized holes. Typically, droplets between about 4 and 40% of the diameterof the hole can be reliably analyzed. Thus a tube with a hole of 16 µm can be used to analyzedroplets with diameters between about 0.6 and 6 µm.

There are a number of practical problems associated with this technique which may limitits application to certain systems. First, it is necessary to dilute an emulsion considerablybefore analysis so that only one particle passes through the hole at a time; otherwise, a pairof particles would be counted as a single larger particle. As mentioned earlier, dilution ofan emulsion may alter the structure of any flocs present, especially when the attractionbetween the droplets is small. Second, the emulsion droplets must be suspended in anelectrolyte solution (typically 5 wt% salt) to ensure that the electrical conductivity of theaqueous phase is sufficiently large to obtain accurate measurements. The presence of anelectrolyte alters the nature of the colloidal interactions between charged droplets, whichmay change the extent of flocculation from that which was present in the original sample.Third, any weakly flocculated droplets may be disrupted by the shear forces generated bythe flocs as they pass through the small hole. Fourth, if an emulsion contains a wide dropletsize distribution, it is necessary to use a number of glass tubes with different hole sizes tocover the full size distribution. Nevertheless, in those emulsions where it can be applied,electrical pulse counting is usually considered to be more reliable and accurate than lightscattering.

10.3.5. Sedimentation Techniques

Sedimentation techniques can be used to determine the size distribution of particles betweenabout 1 nm and 1 mm, although a number of different types of instruments have to be used


to cover the whole of this range (Hunter 1986, 1993). Particle size is determined by measur-ing the velocity at which droplets sediment (or cream) in a gravitational or centrifugal field.

10.3.5.1. Gravitational Sedimentation

The velocity that an isolated rigid spherical particle suspended in a Newtonian liquid movesdue to gravity is given by Stokes’ equation:

vgr

=−2

92 1

2

1

( )ρ ρη (10.6)

where ρ2 is the density of the droplets, ρ1 and η1 are the density and viscosity of thecontinuous phase, g is the acceleration due to gravity, and r is the droplet radius. The dropletsize can therefore be determined by measuring the velocity at which the particle movesthrough the liquid once the densities of both phases and the viscosity of the continuous phaseare known. The movement of the droplets could be monitored using a variety of experimentalmethods, including visual observation, optical microscopy, light scattering, nuclear magneticresonance, ultrasound, and electrical measurements (Mikula 1992, Pal 1994, Dickinson andMcClements 1995). The Stokes equation cannot be used to estimate droplet sizes less thanabout 1 µm because of their Brownian motion (Hunter 1986). The gravitational forces actingon a droplet cause it to move in a certain direction, either up or down, depending on theparticle density relative to the continuous phase. On the other hand, Brownian motion tendsto randomize the spatial distribution of the particles. This effect is important when thedistance a particle moves due to Brownian motion is comparable to the distance it moves dueto sedimentation or creaming in the same time. In a quiescent system, the average distance(root mean square displacement, xrms) that a sphere moves due to Brownian motion is givenby (Hunter 1993):

xkTt

rrms =3πη (10.7)

Thus, a 1-µm oil droplet (ρ2 = 920 kg m–3) moves about 0.7 µm s–1 due to Brownian motionwhen suspended in pure water, whereas it will move about 2 µm in the same time due tocreaming. Thus, even for a droplet of this size, the thermal motion can have a pronouncedinfluence on its creaming rate. In addition, temperature gradients within a sample can leadto convective currents that interfere with the droplet movement. For these reasons, gravita-tional sedimentation techniques have limited application for determining droplet sizes inemulsions. Instead, the droplets are made to move more rapidly by applying a centrifugalforce, so that the problems associated with Brownian motion and convection are overcome.

10.3.5.2. Centrifugation

When an emulsion is placed in a centrifuge and rotated rapidly, it is subjected to a centrifugalforce that causes the droplets to move inward when they have a lower density than thesurrounding liquid (e.g., oil-in-water emulsions) or outward when they have a higher density(e.g., water-in-oil emulsions). The velocity, v(x), at which the droplets move through thesurrounding liquid depends on the angular velocity (ω) at which the tube is centrifuged andtheir distance from the center of the rotor (x) The droplet motion can be convenientlycharacterized by a sedimentation coefficient, which is independent of the angular velocity andlocation of the droplets (Hunter 1993):


Sv x

x= ( )

ω2 (10.8)

The radius of an isolated spherical particle in a fluid is related to the sedimentation coefficientby the following equation:

rS

=−

9

21

2 1

ηρ ρ( ) (10.9)

By measuring the droplet velocity at different positions within a centrifuge tube, it is possibleto determine S and therefore the droplet radius.

The sedimentation coefficient is related to the distance from the rotor center: ln(x2/x1) =Sω2(t2 – t1), where x is the position of the particle at time t (Hunter 1993). Thus a plot of ln(x)versus time at a fixed rotation speed can be used to determine S. For oil-in-water emulsionswith sizes between 0.1 and 100 µm, S varies between about 2 ns and 2 ms. The position ofthe droplets is usually determined by optical microscopy or light scattering. Modern instru-ments are capable of measuring the full droplet size distribution of emulsions by analyzingthe velocity at which a number of different droplets move. Instruments that utilize thisprinciple are commercially available and are used in the food industry.

10.3.6. Ultrasonic Spectrometry


Ultrasonic spectrometry utilizes interactions between ultrasonic waves and particles to obtaininformation about the droplet size distribution of an emulsion (McClements 1991, 1996;Dukhin and Goetz 1996; Povey 1995, 1997). It can be used to determine droplet sizesbetween about 10 nm and 1000 µm. Particle-sizing instruments based on ultrasonic spectrom-etry have recently become available commercially and are likely to gain wide acceptance inthe food industry in the near future because they have a number of important advantages overalternative technologies (e.g., they can be used to analyze emulsions which are concentratedand optically opaque, without the need for any sample preparation).

As an ultrasonic wave propagates through an emulsion, its velocity and attenuation arealtered due to its interaction with the droplets. These interactions may take a number ofdifferent forms: (1) some of the wave is scattered into directions which are different from thatof the incident wave; (2) some of the ultrasonic energy is converted into heat due to variousabsorption mechanisms (e.g., thermal conduction and viscous drag); and (3) there is interfer-ence between waves which travel through the droplets, waves which travel through thesurrounding medium, and waves which are scattered. The relative importance of thesedifferent mechanisms depends on the thermophysical properties of the component phases, thefrequency of the ultrasonic wave, and the concentration and size of the droplets.

As with light scattering, the scattering of ultrasound by a droplet can be divided into threeregimes according to the relationship between the particle size and the wavelength of theradiation: (1) the long-wavelength regime (r < λ/20), (2) the intermediate-wavelength regime(λ/20 < r < 20λ), and (3) the short-wavelength regime (r > 20λ). The wavelength ofultrasound (10 µm to 10 mm) is much greater than that of light (0.2 to 1 µm), and soultrasonic measurements are usually made in the long-wavelength regime, whereas light-scattering measurements are made in the intermediate-wavelength regime. As a consequence,light-scattering results are often more difficult to interpret reliably because of the greatercomplexity of the scattering theory in the intermediate-wavelength regime.


FIGURE 10.15 Dependence of the ultrasonic properties of an emulsion on its droplet size anddispersed-phase volume fraction.

In the long-wavelength regime, the ultrasonic properties of fairly dilute emulsions (φ <15%) can be related to their physicochemical characteristics using the following equation:

Kk

iA

k r

iA

k r1

20

13 3

1

13 3

13

19

= −

−

φ φ(10.10)

where K is the complex propagation constant of the emulsion (= ω/ce + iαe), k1 is the complexpropagation constant of the continuous phase (= ω/c1 + iα1), ω is the angular frequency, cis the ultrasonic velocity, α is the attenuation coefficient, i = √–1, and r is the droplet radius.The A0 and A1 terms are the monopole and dipole scattering coefficients of the individualdroplets, which depend on the adiabatic compressibility, density, specific heat capacity,thermal conductivity, cubical expansivity, and viscosity of the component phases, as well asthe frequency and droplet size (Epstein and Carhart 1953). Equation 10.10 can be used tocalculate the ultrasonic velocity and attenuation coefficient of an emulsion as a function ofdroplet size and concentration, once the thermophysical properties of the component phasesare known, since ce = ω/Re(Ke) and αe = Im(Ke).

The dependence of the ultrasonic velocity and attenuation coefficient of an emulsion ondroplet size and concentration is shown in Figure 10.15. The ultrasonic velocity increaseswith droplet size, while the attenuation per wavelength (αλ) has a maximum value at anintermediate droplet size. It is this dependence of the ultrasonic properties of an emulsion ondroplet size that enables ultrasound to be used as a particle-sizing technology. An emulsionis analyzed by measuring its ultrasonic velocity and/or attenuation spectra and finding thedroplet size and concentration which give the best fit between the experimental data andEquation 10.10. For polydisperse emulsions, the equation has to be modified to take intoaccount the droplet size distribution (McClements 1991). One of the major limitations of theultrasonic technique is the fact that a great deal of information about the thermophysicalproperties of the component phases is needed to interpret the measurements, and this infor-mation often is not readily available in the literature (Coupland and McClements 1998). Inaddition, for droplet concentrations greater than about 15%, it is necessary to extend theabove equation to take into account interactions between the droplets, which has been donerecently (Hemar et al. 1997, McClements et al. 1998b).


The ultrasonic properties of materials can be measured using a number of different experi-mental techniques (e.g., pulse echo, through transmission, and interferometric) (McClements


FIGURE 10.16 Ultrasonic pulse-echo measurement technique.

1996). The major difference between them is the form in which the ultrasonic energy isapplied to the sample and the experimental configuration used to carry out the measurements.In this section, we shall only consider the pulse-echo technique because it is one of thesimplest and most widely used techniques to analyze emulsions. The sample to be analyzedis contained within a thermostated measurement cell which has an ultrasonic transducer fixedto one side (Figure 10.16). An electrical pulse is applied to the transducer, which stimulatesit to generate a pulse of ultrasound that is directed into the sample. This pulse travels acrossthe sample, is reflected from the back wall of the measurement cell, travels back through thesample, and is then detected by the same transducer (Figure 10.16). The ultrasonic pulsereceived by the transducer is converted back into an electrical pulse which is digitized andsaved for analysis.

The ultrasonic velocity and attenuation coefficient of a sample are determined by measur-ing the time of flight (t) and amplitude (A) of the ultrasonic pulse which has traveled throughit. The ultrasonic velocity is equal to the distance traveled by the pulse (2d) divided by itstime of flight: c = 2d/t. The attenuation coefficient is calculated by comparing the amplitudeof a pulse which has traveled through the sample with that of a pulse which has traveledthrough a material whose attenuation coefficient is known: α2 = α1 – ln(A2 /A1)/2d, where thesubscripts 1 and 2 refer to the properties of the reference material and the material beingtested, respectively. The through-transmission technique is very similar to the pulse-echotechnique, except that separate transducers are used to generate and receive the ultrasonicpulse (McClements 1996).


To determine the droplet size distribution of an emulsion, it is necessary to measure thefrequency dependence of its ultrasonic properties. Two different approaches can be used tomeasure the dependence of the ultrasonic properties on frequency using the pulse-echotechnique. In the first approach, a broadband ultrasonic pulse is used, which is a single pulsethat contains a wide range of frequencies. After the pulse has traveled through the sample,it is analyzed by a Fourier transform algorithm to determine the values of t and A (andtherefore c and α) as a function of frequency (McClements and Fairley 1991, 1992). To coverthe whole frequency range (0.1 to 100 MHz), a number of transducers with different frequen-cies have to be used. In the second approach, a tone-burst ultrasonic pulse is used, which isa single pulse that contains a number of cycles of ultrasound at a particular frequency(McClements 1996). The transducer is tuned to a particular frequency, and a measurementof the ultrasonic velocity and attenuation coefficient is made. The transducer is then tuned toanother frequency and the process is repeated. Because measurements are carried out sepa-rately at a number of different frequencies, this approach is more time consuming andlaborious than the one which uses broadband pulses. Both of these methods are used todetermine the frequency dependence of the ultrasonic properties of emulsions in commercialultrasonic particle sizers.


Ultrasound has major advantages over many other particle-sizing technologies because it canbe used to measure droplet size distributions in concentrated and optically opaque emulsionsin situ. In addition, it can be used as an on-line sensor for monitoring the characteristics offood emulsions during processing, which gives food manufacturers much greater control overthe quality of the final product. The possibility of using ultrasound to measure droplet sizesin real foods has been demonstrated for casein micelles (Griffin and Griffin 1990), sunflower-oil-in-water emulsions (McClements and Povey 1989), corn-oil-in-water emulsions (Couplandand McClements 1998), salad creams (McClements et al. 1990), and milk fat globules (Mileset al. 1990).

The two major disadvantages of the ultrasonic technique are the large amount ofthermophysical data required to interpret the measurements and the fact that small air bubblescan interfere with the signal from the emulsion droplets (McClements 1991, 1996).

10.3.7. Nuclear Magnetic Resonance

Instrumental techniques based on nuclear magnetic resonance (NMR) utilize interactionsbetween radio waves and the nuclei of hydrogen atoms to obtain information about theproperties of materials.* An NMR technique has been developed to measure the droplet sizedistribution of emulsions (Callaghan et al. 1983, van den Enden et al. 1990, Li et al. 1992,Soderman et al. 1992), which is sensitive to particle sizes between about 0.2 and 100 µm(Dickinson and McClements 1995). This technique relies on measurements of the restricteddiffusion of molecules within emulsion droplets.

The principles of the technique are fairly complex and have been described in detailelsewhere (Dickinson and McClements 1995, Soderman and Bailnov 1996). Basically, thesample to be analyzed is placed in a static magnetic field gradient and a series of radio-frequency pulses are applied to it. These pulses cause some of the hydrogen nuclei in thesample to be excited to higher energy levels, which leads to the generation of a detectableNMR signal. The amplitude of this signal depends on the movement of the nuclei in thesample: the farther the nuclei move during the experiment, the greater the reduction in the

* NMR techniques can also be used to study the nuclei of certain other isotopes, but these are not widely used forparticle sizing.


amplitude. A measurement of the signal amplitude can therefore be used to study molecularmotion.

In a bulk liquid, the distance that a molecule can move in a certain time is governed by itstranslational diffusion coefficient, xrms = (√2Dt). When a liquid is contained within anemulsion droplet, its diffusion may be restricted because of the presence of the interfacialboundary. If the movement of a molecule in a droplet is observed over relatively short times(t << d2/2D), the diffusion is unrestricted, but if it is observed over longer times, its diffusionis restricted because it cannot move farther than the diameter of the droplet. By measuringthe attenuation of the NMR signal at different times, it is possible to identify when thediffusion becomes restricted and thus estimate the droplet size. Because this technique relieson the movement of molecules within droplets, it is independent of droplet flocculation.

This technique has been used to determine the droplet size distribution of a variety of oil-in-water and water-in-oil emulsions, including margarine, cream, and cheese (Callaghan et al.1983, van den Enden et al. 1990, Li et al. 1992, Soderman et al. 1992). Like ultrasonicspectrometry, it is nondestructive and can be used to analyze emulsions which are concen-trated and optically opaque (Dickinson and McClements 1995). It therefore seems likely thatNMR instruments specifically designed to measure the particle size distribution of emulsionswill be developed and become commercially available in the near future.

10.3.8. Neutron Scattering

Neutron scattering techniques utilize interactions between a beam of neutrons and an emul-sion to determine the droplet size distribution (Dickinson and Stainsby 1982, Eastoe 1995).They can also be used to provide information about the thickness of interfacial layers and thespatial distribution of droplets. These techniques have a couple of special features whichmake them particularly suitable for studying food emulsions (Eastoe 1995). First, the scat-tering of neutrons from emulsion droplets is very weak, and therefore multiple scatteringeffects are not appreciable, which means that concentrated emulsions can be analyzed withoutdilution. Second, the scattering of neutrons from heterogeneous materials depends on the“contrast” between the different components, which can be manipulated by the experimenter.Thus it is possible to selectively highlight specific structural features within an emulsion (seebelow). Despite its ability to generate information which is difficult to obtain using othertechniques, the application of neutron scattering to food emulsions is limited because anuclear reactor is needed to generate the neutron beam. There are only a small number ofneutron-scattering facilities in the world which are generally accessible, and beam time israther limited and must be scheduled many months in advance of the proposed experiment(Stothart 1995).

In many respects, the measurement principle of neutron scattering is similar to that of staticlight scattering, except that a beam of neutrons is used instead of light. The sample to beanalyzed is placed into a cuvette which is inserted between a source of neutrons and a neutrondetector (Stothart 1995). A beam of neutrons is passed through the emulsion, and the intensityof the scattered neutrons is measured as a function of scattering angle (and/or wavelength).Information about the properties of the emulsion is then obtained by interpreting the resultingspectra using an appropriate neutron-scattering theory (Eastoe 1995).

Each type of atomic nuclei scatters neutrons to a different extent, which is characterizedby a “scattering cross-section” (Lovsey 1984). The scattering of neutrons from a heterog-enous material, such as an emulsion, depends on the contrast between the scattering cross-sections of the different components: the greater the contrast, the more intense the scattering.One of the most important attributes of neutron scattering is the ability to alter the scatteringcross-section of molecules that contain hydrogen atoms (e.g., water, proteins, fats, andcarbohydrates) (Eastoe 1995, Stothart 1995). Normal hydrogen (1H) and deuterium (2H) have


FIGURE 10.17 Concept of contrast matching of emulsions in neutron-scattering experiments.

significantly different scattering cross-sections, and so by varying the 1H:2H ratio of aparticular type of molecule, it is possible to increase or decrease its contrast with respect tothe other components in an emulsion. As a consequence, it is possible to emphasize specificstructural components within an emulsion.

Consider an oil-in-water emulsion (Figure 10.17) which consists of oil droplets covered byan interfacial layer and suspended in an aqueous phase. By altering the ratio of water todeuterated water in the aqueous phase, it is possible to match the aqueous phase to either theinterfacial layer or the oil within the droplets. Alternatively, the interfacial layer could bematched to the droplet by partial deuteration of either the oil or emulsifier molecules. Thusit is possible to obtain information about the thickness of the interfacial layer, the oil droplet,or the droplet + interfacial layer.

10.3.9. Dielectric Spectroscopy

This technique depends on the dielectric response of an emulsion during the application ofan electromagnetic wave (Clausse 1983, Asami 1995, Sjoblom et al. 1996). The possibilityof using dielectric spectroscopy to determine the droplet size distribution of concentratedemulsions was recently demonstrated (Garrouch et al. 1996). The dielectric permittivity of anemulsion is measured over a wide range of electromagnetic frequencies, and the resultingspectra are analyzed using a suitable theory to determine the droplet size distribution. Themajor limitation of this technique is that it can only be used to determine droplet sizedistributions in emulsions that contain charged particles. Even so, it can be used to simulta-neously measure the zeta potential and droplet size distribution, and it can be used to analyzeemulsions which are concentrated and optically opaque without the need for any sampledilution. It may therefore have some important applications in the food industry. Neverthe-less, dielectric spectroscopy is still in its infancy, and a lot more research is still requiredbefore the technique becomes more widely accepted and utilized.


The droplet size distribution of emulsions that contain charged droplets can be determinedusing electroacoustic techniques (O’Brien et al. 1995, Carasso et al. 1995, Dukhin and Goetz1996). These techniques use a combination of electric and acoustic phenomena to determineboth the size and zeta potential of emulsion droplets (Section 10.6.3). Instruments based onthis principle have recently become commercially available and are capable of analyzingdroplets with sizes between about 0.1 and 10 µm (Hunter 1998). These instruments arecapable of analyzing emulsions with high droplet concentrations (<40%) without any sampledilution. The major limitation of the electroacoustic technique is that measurements rely on


the droplets having an electrical charge, the surrounding liquid must be Newtonian, and theremust be a significant density contrast between the droplets and the surrounding liquid. Theseconditions are not always met for food emulsions, which means that electroacoustics mayonly have limited application within the food industry.

10.4. DISPERSED-PHASE VOLUME FRACTION

10.4.1. Proximate Analysis

The concentration of droplets in an emulsion can be determined using many of the standardanalytical methods developed to determine the composition of foods (Nielsen 1994, Pomeranzand Meloan 1994). A variety of solvent-extraction techniques are available for measuringfat content (Min 1994, Pal 1994). The sample to be analyzed is mixed with a nonpolarorganic solvent which extracts the oil. The solvent is then physically separated from theaqueous phase, and the oil content is determined by evaporating the solvent and weighingthe residual oil. A possible difficulty associated with applying this technique to oil-in-wateremulsions is that the interfacial membrane may be resistant to rupture, and therefore all ofthe oil is not released. This problem has been overcome in a number of nonsolvent tech-niques developed to measure the fat content of dairy emulsions. In the Gerber and Babcockmethods, an emulsion is placed in a specially designed bottle and then mixed with sulfuricacid, which digests the interfacial membrane surrounding the droplets and thus causescoalescence (Min 1994). The bottle is centrifuged to facilitate the separation of the oil andaqueous phases, and the percentage of oil in the emulsion is determined from the calibratedneck of the bottle. A similar procedure is involved in the detergent method, except thatrather than sulfuric acid, a surfactant is added to promote droplet coalescence.

The water content of an emulsion can be determined using a variety of proximate analysistechniques (Mikula 1992, Pal 1994, Pomeranz and Meloan 1994, Bradley 1994). The sim-plest of these involves weighing an emulsion before and after the water has been evaporated,which may be achieved by conventional oven, vacuum oven, microwave oven, or infraredlight. The moisture content can also be determined by distillation. The emulsion is placed ina specially designed flask which has a calibrated side arm. An organic solvent is mixed withthe emulsion, and the flask is heated to cause the water to evaporate and collect in the sidearm. This procedure is continued until all of the water has evaporated, and then its volumeis determined from the calibrations on the side arm.

Many of these techniques are labor intensive, time consuming, and destructive andtherefore are unsuitable for rapid quality assurance tests. Instruments based on infraredabsorption are becoming increasingly popular for rapid and nondestructive analysis of foodcomposition (Wehling 1994, Wilson 1995). Once calibrated, these instruments are capableof simultaneously determining the concentration of fat, water, protein, and carbohydrate.These instruments are likely to find widespread use in the food industry, particularly for on-line measurements.

10.4.2. Density Measurements


One of the simplest methods of determining the dispersed-phase volume fraction of anemulsion is to measure its density (Pal 1994). The density of an emulsion (ρe) is related tothe densities of the continuous (ρ1) and dispersed phases (ρ2): ρe = φρ2 + (1 – φ)ρ1. Thedispersed-phase volume fraction of an emulsion can therefore be determined by measuringits density and knowing the density of the oil and aqueous phases:


φρ ρρ ρ

=−−

e 1

2 1(10.11)

The densities of the dispersed and continuous phases in a food emulsion are appreciablydifferent, about 900 kg m–3 for liquid oils and about 1000 kg m–3 for aqueous phases. It ispossible to measure the density of an emulsion to about 0.2 kg m–3, and so the dispersed-phase volume fraction can be determined to within 0.002 (0.2%) using this technique. Thefact that the density of liquid oils is lower than that of water means that the density of anemulsion usually decreases with increasing oil content (Figure 10.18).


Density Bottles. A liquid sample is poured into a glass bottle of known mass and volume(Pomeranz and Meloan 1994). The bottle and liquid are allowed to equilibrate to the measure-ment temperature and are then weighed using an accurate balance. The mass of emulsion(memulsion) required to completely fill the container at a given temperature is measured. Theinternal volume of the container is determined by measuring the mass of distilled water (amaterial whose density is known accurately) it takes to fill the bottle (Vbottle = mwater/ρwater).Thus, the density of the emulsion can be determined: ρemulsion = memulsion/Vbottle. The densityof an emulsion is particularly sensitive to temperature, and so it is important to carefullycontrol the temperature when accurate measurements are required. It is important to ensurethat the container is clean and dry prior to weighing and that it contains no gas bubbles. Gasbubbles reduce the volume of liquid in the bottle without contributing to the mass andtherefore lead to an underestimate of the density.

Hydrometers. Several methods for measuring the density of liquids are based on theArchimedes principle, which states that the upward buoyant force exerted on a body im-

FIGURE 10.18 Dependence of the physicochemical properties of various oil-in-water emulsions ontheir dispersed-phase volume fraction.


mersed in a liquid is equal to the weight of the displaced liquid (Pomeranz and Meloan 1994).This principle is used in hydrometers, which are graduated hollow glass bodies that float onthe liquid to be tested. The depth that the hydrometer sinks into a liquid depends on thedensity of the liquid. The hydrometer sinks to a point where the mass of the displaced liquidis equal to the mass of the hydrometer. Thus the depth to which a hydrometer sinks increasesas the density of the liquid decreases. The density of a liquid is read directly from graduatedcalibrations on the neck of the hydrometer. Hydrometers are less accurate than densitybottles, but much more rapid and convenient to use.

Oscillating U-tubes. The density of fluids can be measured rapidly and accurately using aninstrument called an oscillating U-tube densitometer (Pal 1994). The sample to be analyzedis placed in a glass U-tube, which is forced to oscillate sinusoidally by the application of analternating mechanical force. The resonant frequency of the U-tube is related to its mass andtherefore depends on the density of the material contained within it. The density of a fluidis determined by measuring the resonant frequency of the U-tube and relating it to the densityusing an appropriate mathematical equation. The instrument must be calibrated with twofluids of accurately known density (usually distilled water and air). Density can be measuredto within 0.1 kg m–3 in a few minutes using this technique. Recently, on-line versions of thistechnique have been developed for monitoring the density of fluids during processing. Asmall portion of a fluid flowing through a pipe is directed through an oscillating U-tube andits density is measured before being redirected into the main flow.


An accurate measurement of emulsion density can be carried out using inexpensive equip-ment that is available in many laboratories. The technique is nondestructive and can be usedto analyze emulsions which are concentrated and optically opaque. One possible problemwith the technique is that the physical state of the emulsion constituents may alter theaccuracy of a measurement. For example, the density of solid fat is greater than that of liquidoil, and therefore the density of an emulsion depends on the solid fat content, as well as thetotal fat content. In these situations, it is necessary to heat the emulsion to a temperaturewhere it is known that all of the fat crystals have melted and then measure its density.

10.4.3. Electrical Conductivity


The dispersed-phase volume fraction of an emulsion can be conveniently determined bymeasuring its electrical conductivity (ε) (Clausse 1983, Robin et al. 1994, Asami 1995). Theelectrical conductivity of water is much higher than that of oil, and so there is a decrease inε as the oil content of an emulsion increases (Figure 10.18). In dilute emulsions, the dis-persed-phase volume fraction is related to the electrical conductivity by the following equa-tion (Clausse 1983):

φε εε ε

ε εε ε

=−+

+−

e

e

1

1

2 1

2 12

2(10.12)

where the subscripts 1, 2, and e refer to the continuous phase, dispersed phase, and emulsion,respectively. More complex expressions have been derived to relate the dispersed-phasevolume fraction of concentrated emulsions to their electrical properties.



The electrical conductivity of an emulsion can simply be determined using a conductivity cell(Siano 1998). This cell consists of a couple of electrodes which are connected to electricalcircuitry that is capable of measuring the electrical conductivity of the sample containedbetween the electrodes. The electrical conductivity of an aqueous phase depends on theconcentration of electrolytes present, and so it is important to properly characterize theproperties of the component phases.


The electrical conductivity technique can be used to determine the dispersed-phase volumefraction of concentrated and optically opaque emulsions without the need for any samplepreparation. Measurements are independent of the size of the emulsion droplets, which is anadvantage when the droplet size distribution is unknown (Robin et al. 1994). The electricalconductivity of an emulsion is dependent on the physical state of its constituents, andtherefore it may be necessary to heat an emulsion to a temperature where all of the crystalshave melted before making a measurement. The electrical conductivity is also sensitive to theionic strength of the aqueous phase, and therefore it is necessary to take this into accountwhen carrying out the analysis (Skodvin et al. 1994).

10.4.4. Alternative Techniques

The dispersed-phase volume fraction can be measured using many of the techniques used todetermine droplet size distributions (Section 10.3). Light-scattering and electrical pulse countingtechniques can be used to determine dispersed-phase volume fractions in dilute emulsions (φ< 0.1%), whereas Doppler shift spectroscopy, ultrasonic, electroacoustic, dielectric, neutron-scattering, and NMR techniques can be used to analyze much more concentrated emulsions.All of these techniques rely on there being a measurable change in some physicochemicalproperty of an emulsion as its droplet concentration increases (e.g., the intensity of scatteredor transmitted light, the attenuation or velocity of an ultrasonic wave, the amplitude or decaytime of an NMR signal) (Figure 10.18). Some of these techniques can be used to simulta-neously determine the droplet size distribution and dispersed-phase volume fraction, whereasothers can be used to determine φ independently of a knowledge of the droplet size.

10.5. DROPLET CRYSTALLINITY

10.5.1. Dilatometry


Dilatometry has been used for many years to monitor the crystallinity of both the dispersedand continuous phases in emulsions (Turnbull and Cormia 1961, Skoda and van den Tempel1963, Phipps 1964). The technique is based on measurements of the density change whichoccurs when a material melts or crystallizes. The density of the solid state of a material isusually greater than that of the liquid state because the molecules are able to pack moreefficiently.* Consequently, there is a decrease in the density of a material when it melts andan increase when it crystallizes. The fraction of crystalline droplets in an emulsion can bedetermined from the following equation:

* With the important exception of water near its freezing point.


φρ ρρ ρc

e eL

eS eL

=−− (10.13)

where, ρe is the density of an emulsion that contains partially crystalline droplets, and ρeL andρeS are the densities of the same emulsion when the droplets are either completely liquid orcompletely solid, respectively. The values of ρeL and ρeS are usually determined by extrapo-lating density measurements from higher and lower temperatures into the region where thedroplets are partially crystalline.


In principle, dilatometry can be carried out using any experimental technique that is capableof measuring density changes. In practice, dilatometry is often performed using a speciallydesigned piece of apparatus (Figure 10.19). A known mass of sample is placed into a glassbulb which is connected to a calibrated capillary tube. A liquid, such as mercury or coloredwater, is poured into the capillary tube above the fat. The change in volume of the samplewhen it crystallizes or melts is then determined by observing the change in height of the liquidin the capillary tube. These measurements can be carried out either as the temperature of thesample is varied in a controlled way or as the sample is held at a constant temperature overtime (Turnbull and Cormia 1961). The data can be presented as a volume change or as adensity change, depending on which is most convenient.


The temperature dependence of the density of an oil-in-water emulsion that contains drop-lets which undergo a phase transition is shown in Figure 10.20. When the emulsion is heatedfrom a temperature where the droplets are completely solid, there is a sharp decrease indensity when the droplets melt. Conversely, when an emulsion is cooled from a tempera-ture where the droplets are completely liquid, there is a sudden increase in density whenthe droplets crystallize. The crystallization temperature is considerably lower than themelting temperature because of supercooling (Section 4.2). In food emulsions, oil dropletsnormally melt over a much wider temperature range than that shown for a pure oil in Figure10.20 because they contain a mixture of different triacylglycerols, each with its own meltingpoint.

FIGURE 10.19 Schematic diagram of a simple dilatometer used for monitoring phase transitions ofmaterials.


FIGURE 10.20 Temperature dependence of the density of a hexadecane oil-in-water emulsion. Forfood oils, melting and crystallization normally occur over a wider range of temperatures.

10.5.2. Nuclear Magnetic Resonance


NMR has been used to measure the solid content of food emulsions for many years (Walstraand van Beresteyn 1975, van Boekel 1981, Dickinson and McClements 1995). NMR instru-ments are capable of rapidly analyzing emulsions that are concentrated and optically opaque,without the need for any sample preparation. For this reason, they have largely replaced themore cumbersome and time-consuming dilatometry method in laboratories which can affordthe relatively high initial cost of an NMR instrument. The NMR technique utilizes interac-tions between radio waves and the nuclei of hydrogen atoms to obtain information about thesolid content of a material. The fundamental principles of this application have been de-scribed elsewhere (Dickinson and McClements 1995), and so only a simplified descriptionof the technique will be given here. Basically, a radio-frequency pulse is applied to anemulsion, which causes some of the hydrogen nuclei to move into an excited state, whichleads to the generation of a detectable NMR signal. The frequency, amplitude, and decay timeof this signal depend on the ratio of solid to liquid material in the sample. Thus, by analyzingthe characteristics of the NMR signal, it is possible to obtain information about the solidcontent of the material.


A variety of NMR instruments which can be used to measure the solid content of emulsionsare commercially available. These instruments vary in their operating principles, the types ofinformation they are capable of providing, and their cost. The more sophisticated instrumentscan measure a large number of different physicochemical characteristics of emulsions, butbecause they are often very expensive and require highly trained operators, their applicationis restricted to a small number of research laboratories. A number of less sophisticatedinstruments are available that have a more limited range of applications but which areconsiderably less expensive. These instruments are the most widely used to determine thesolid content of foods, and therefore only they will be considered here.

A typical pulsed NMR instrument consists of a static magnet, a probe coil, electronics togenerate and receive electromagnetic pulses, and a computer to regulate the measurementprocedure and analyze and store the data (Figure 10.21). Nuclei in the sample are excited to


FIGURE 10.22 Decay of NMR signal after the application of an NMR pulse is much more rapid forsolid than liquid fat.

FIGURE 10.21 A typical arrangement for making NMR measurements of the solid content of anemulsion.

a higher energy level by applying a radio-frequency pulse via the probe coil, and the excitednuclei are detected by the same coil once the radio-frequency pulse is removed. The detectedsignal is digitized by a digital-to-analog convertor and stored in the computer for dataanalysis. The solid content is determined by analyzing the decay rate of the detected signalafter the application of the radio-frequency pulse.

The decay of the NMR signal is much more rapid for a solid than a liquid (Figure 10.22).Consequently, the solid and liquid phases in a sample can be differentiated by measuring thedecay of the NMR signal with time. Immediately after the radio-frequency pulse is switchedoff, the NMR signal (S0) is proportional to the total number of nuclei in the liquid and solidphases. After a certain time, the contribution from the solid phase has completely decayed,and the signal (St) is then proportional to the number of nuclei in the liquid phase. Assumingthat the NMR signal per gram of the solid and liquid phases is similar, the solid content cansimply be determined: φc = (S0 – St)/S0. In practice, it is not possible to measure the signalat zero time because the NMR receiver takes a short time to recover after the radio-frequency


pulse is applied. Consequently, it is necessary to use a correction factor that accounts for theslight decay of the signal before the first measurement is made. The solid content can alsobe determined by only making measurements of the signal from the liquid phase, since St isproportional to the mass of liquid phase present (ML). The mass of liquid in a sample isdeduced by measuring St and using a previously prepared calibration curve of ML versus St.If the total mass of the sample analyzed (MT) is known, then φc = (MT – ML)/MT. The solidcontent can also be determined by measuring St for a partially crystalline sample and thenheating it to a temperature where all of the solid phase melts and measuring it again: φc = (S′t– St)/S′t, where the prime refers to the measurement at the higher temperature. This value hasto be corrected to take into account the temperature dependence of the NMR signal of theliquid phase.


NMR has been used to determine the extent of droplet crystallization in emulsions in anumber of fundamental studies and commercial products (Walstra and van Beresteyn 1975,Waddington 1980, van Boekel 1981, Dickinson and McClements 1995). Benchtop instru-ments are available which are extremely simple and rapid to use. A tube containing thesample is placed in the NMR instrument and the solid content is given out in a few minutesor less. The technique is therefore particularly useful for quality control purposes when manysamples have to be rapidly analyzed. Attempts have been made to develop on-line versionsof these NMR instruments, but their application is limited because the sample cannot beanalyzed within a metal pipe because of its distorting influence on the magnetic field. Anadditional limitation is that the technique cannot be used to accurately determine the solidcontents when the degree of crystallization is small (<5%).

Before leaving this section, it should be noted that there are a variety of other NMRtechniques which can also be used to determine droplet solid contents (Dickinson andMcClements 1995). The most powerful of these is NMR imaging, which can be used todetermine the solid content at any location within a material, so that it is possible to obtaina three-dimensional image of the droplet crystallinity (Simoneau et al. 1991, 1993). Theseimaging techniques provide food scientists with an extremely powerful method of monitoringand predicting the long-term stability of food emulsions, which will undoubtedly help toidentify the most important factors that determine emulsion properties. Nevertheless, thesetechniques are extremely expensive and require highly skilled operators and therefore arecurrently available to a small number of research laboratories.

10.5.3. Thermal Analysis


Differential thermal analysis (DTA) and differential scanning calorimetry (DSC) are twothermal analysis techniques that can be used to monitor melting and crystallization of emul-sion droplets (Walstra and van Beresteyn 1975, McClements et al. 1993a, Davis 1994a).These techniques are based on measurements of the heat released or adsorbed by a samplewhen it is subjected to a controlled temperature program. A material tends to release heatwhen it crystallizes and absorb heat when it melts. The major difference between the twotechniques is the method used to measure the heat adsorbed or released by the sample.


Differential Thermal Analysis. DTA records the difference in temperature between a sub-stance and a reference material when they are heated or cooled at a controlled rate. A DTA


FIGURE 10.23 DTA and DSC instruments.

instrument consists of two sample holders which are connected to the same furnace, whichis a device for varying the temperature in a controlled fashion (Figure 10.23). A few milli-grams of sample is accurately weighed into a small aluminum pan and placed into one of thesample holders. A reference, usually an empty aluminum pan or a pan containing a materialthat does not undergo a phase transition over the temperature range studied, is placed into theother sample holder. The sample and reference pans are then heated or cooled together at acontrolled rate. The difference in temperature (∆T) between the two pans is recorded bythermocouples located below each of the pans. The output from the instrument is thereforea plot of ∆T versus T. If the sample absorbs heat (an endothermic process), then it will havea slightly lower temperature than the reference, but if it releases heat (an exothermic process),it will have a slightly higher temperature. Thus measurements of the difference in tempera-ture between the sample and reference pans can be used to provide information about physicaland chemical changes which occur in the sample. An exothermic process leads to thegeneration of a negative peak, and an endothermic process leads to the generation of apositive peak (Figure 10.24).

Differential Scanning Calorimetry. DSC records the energy necessary to establish a zerotemperature difference between a sample and a reference material which are either heated orcooled at a controlled rate. Thermocouples constantly measure the temperature of each pan,and two heaters below the pans supply heat to one of the pans so that they both have exactlythe same temperature. If a sample were to undergo a phase transition, it would either absorbor release heat. To keep the temperature of the two pans the same, an equivalent amount ofenergy must be supplied to either the test or reference cells. Special electrical circuitry is usedto determine the amount of energy needed to keep the two sample pans at the same tempera-ture. DSC data are therefore reported as the rate of energy absorption (Q) by the samplerelative to the reference material as a function of temperature.


Thermal analysis can be used to determine the temperature range of a phase transition, as wellas the amount of material involved in a phase transition (Figure 10.24). When an emulsion


FIGURE 10.24 Thermal analysis of the melting and crystallization of oil droplets in an oil-in-wateremulsion.

that contains solid droplets is heated above the melting point of the dispersed phase, thedroplets melt and an exothermic peak is observed. The melting temperature can therefore beascertained by measuring the position of the peak. The area under the peak is proportionalto the amount of material which undergoes the phase transition: A = k∆Hfm, where ∆Hf is theheat of fusion per gram, m is the mass of material undergoing the phase transition in grams,and k is a constant which depends on the instrument settings used to make the measurement.The value of k is determined by measuring the peak areas of a series of samples of knownmass and heat of fusion. Thus the mass of material which melts can be determined bymeasuring the peak area. When an emulsion that contains liquid droplets is cooled, anendothermic peak is observed when the droplets crystallize. The crystallization temperatureand the amount of material which has crystallized can be determined in the same way as forthe melting curve. As mentioned earlier, the droplets tend to crystallize at a much lowertemperature than they melt because of supercooling. In addition, the melting range of ediblefats is much wider than that shown in Figure 10.24 because they contain a mixture oftriacylglycerols.

Thermal analysis has been used to monitor the influence of oil type, emulsifier type,cooling rates, catalytic impurities, polymorphic changes, and droplet size on the nucleationand crystallization of droplets in oil-in-water emulsions (Dickinson and McClements 1995).It has also been used to study the factors which influence the melting and crystallization ofwater droplets in water-in-oil emulsions (Clausse 1985) and to study phase transitions in thecontinuous phase of emulsions.

10.5.4. Ultrasonics


The ultrasonic properties of a material change significantly when it melts or crystallizes, andso ultrasound can be used to monitor phase transitions in emulsions (Dickinson et al. 1990,1991c; McClements 1991). The temperature dependence of the ultrasonic velocity of an oil-in-water emulsion in which the droplets crystallize is shown in Figure 10.25. As the emulsionis cooled to a temperature where the droplets crystallize, the ultrasonic velocity increasessteeply, because the velocity of ultrasound is larger in solid fat than in liquid oil. As the


emulsion is heated to a temperature where the droplets melt, the velocity decreases rapidly.The droplets crystallize at a temperature which is much lower than the melting point of thebulk oil because of supercooling effects (Section 4.2).

To a first approximation, the fraction of crystalline material (φSFC) in an emulsion can bedetermined using the following equation (Dickinson et al. 1991):

φSFC = −

−

−1 1 1 12 2 2 2

1

c c c ce eL eS eL(10.14)

where ceL and ceS are the ultrasonic velocities in the emulsion if all the droplets were eithercompletely liquid or completely solid, respectively. These values are determined by extrapo-lating measurements from higher and lower temperatures into the region where the fat ispartially crystalline or by using ultrasonic scattering theory to calculate their values.


The ultrasonic properties of an emulsion can be measured using the same techniques as usedfor determining the droplet size distribution (Section 10.3). The measurements are usuallycarried out in one of two ways: isothermal or temperature scanning. In an isothermal experi-ment, the temperature of the emulsion is kept constant and the change in the ultrasonicvelocity is measured as a function of time. In a temperature scanning experiment, theultrasonic velocity is measured as the temperature is increased or decreased at a controlledrate.


The solid contents determined using ultrasound are in good agreement with those determinedusing traditional techniques such as dilatometry (Hussin and Povey 1984) and NMR(McClements and Povey 1988). Ultrasound has been used to monitor phase transitions innonfood oil-in-water emulsions (Dickinson et al. 1990, 1991), triacylglycerol oil-in-water andwater-in-oil emulsions (McClements 1989, Coupland et al. 1993), margarine and butter

FIGURE 10.25 Crystallization and melting of oil droplets in an emulsion can be monitored byultrasonic velocity measurements.


(McClements 1989), shortening and meat (Miles et al. 1985), and various triacylglycerol/oilmixtures (McClements and Povey 1987, 1988).

It should be noted that in some systems, large increases in the attenuation coefficient andappreciable velocity dispersion have been observed during melting and crystallization be-cause of a relaxation mechanism associated with the solid–liquid phase equilibrium(McClements et al. 1998). The ultrasonic wave causes periodic fluctuations in the tempera-ture and pressure of the material, which perturb the phase equilibrium. When a significantproportion of the material is at equilibrium, a large amount of ultrasonic energy is absorbeddue to this process. This effect depends on the ultrasonic frequency, the relaxation time forthe phase equilibrium, and the amount of material undergoing phase equilibrium and cancause large deviations in both the velocity and attenuation coefficient. In systems where thisphenomenon is important, it is not possible to use Equation 10.14 to interpret the data.Nevertheless, it may be possible to use ultrasound to obtain valuable information about thedynamics of the phase equilibrium.

10.6. DROPLET CHARGE

10.6.1. Electrophoresis

The emulsion to be analyzed is placed in a measurement cell, and a static electrical field (E)is applied across it via a pair of electrodes (Figure 10.26). This causes any charged emulsiondroplets to move toward the oppositely charged electrode (Hunter 1986, 1993). The sign ofthe charge on the emulsion droplets can therefore be deduced from the direction they move.When an electrical field is applied across an emulsion, the droplets accelerate until they reacha constant velocity (v) where the electrical pulling force is exactly balanced by the viscousdrag force exerted by the surrounding liquid. This velocity depends on the size and chargeof the emulsion droplets and can therefore be used to provide information about theseparameters. Experimentally, particle velocity is determined by measuring the distance theymove in a known time or the time it takes to move a known distance. Droplet motion can bemonitored using a number of different experimental methods. The movement of relativelylarge particles (>1 µm) can be monitored by optical microscopy or static light scattering,whereas the movement of smaller particles can be monitored by an ultramicroscope ordynamic light scattering.

Mathematical expressions have been derived to relate the movement of a droplet in anelectric field to its zeta potential (Hunter 1986). These are based on a theoretical consider-ation of the forces that act on a particle when it has reached constant velocity (i.e., theelectrical pulling force is balanced by the viscous drag force). The mathematical theory thatdescribes this process depends on the droplet size and charge, the thickness of the Debye

FIGURE 10.26 Electrophoretic mobility cell for measuring the zeta potential of emulsion droplets.


layer, the viscosity of the surrounding liquid, and the strength of the applied electric field. Thegeneral solution of this theory leads to a complicated expression that relates all of theseparameters. Nevertheless, under certain experimental circumstances, it is possible to derivesimpler expressions:

ξ ηε ε

κ= <<32

10

ur

R(10.15)

ξ ηε ε

κ= >>ur

R0

1 (10.16)

where η is the viscosity of the surrounding liquid, u is the electrophoretic mobility (= particlevelocity divided by electric field strength), ε

0 is the dielectric constant of a vacuum, and εR

is the relative dielectric constant of the material. In practice, the latter equation is the mostapplicable to emulsions, because the droplet size is much greater than the Debye length (κ –1).Even so, there are many practical examples where the particle size is comparable to the Debyelength, and so there are considerable deviations between Equation 10.16 and experimentalmeasurements. In these cases, it is necessary to solve the full theory.

10.6.2. Zetasizer©

A more sophisticated instrument for measuring both the zeta potential and size of dropletsin emulsions is the Zetasizer© developed by Malvern Instruments (Hunter 1986). Twocoherent beams of light are made to intersect with each other at a particular position withina measurement cell so that they form an interference pattern which consists of regions of lowand high light intensity. The charged emulsion droplets are made to move through theinterference pattern by applying an electrical field across the cell. As the droplets moveacross the interference pattern, they scatter light in the bright regions, but not in the darkregions. The faster a droplet moves through the interference pattern, the greater the fre-quency of the intensity fluctuations. By measuring and analyzing the frequency of thesefluctuations, it is possible to determine the particle velocity, which can then be mathemati-cally related to the zeta potential (e.g., using Equation 10.16 for larger particles). The signof the charge on the particles is ascertained from the direction they move in the electric field.The same instrument can also be used to determine droplet concentration and size distribu-tion (from 10 nm to 3 µm) of an emulsion by a dynamic light-scattering technique. Conse-quently, it is possible to determine the droplet size, concentration, and charge using a singleinstrument, which is extremely valuable for predicting the stability and bulk physicochemi-cal properties of emulsions.


Recently, analytical instruments based on electroacoustics have become commercially avail-able for measuring the size, concentration, and zeta potential of droplets in emulsions (Hunter1993, O’Brien et al. 1995, Carasso et al. 1995, Dukhin and Goetz 1996). The sample to beanalyzed is placed in a measurement cell and an alternating electrical field is applied acrossit via a pair of electrodes. This causes any charged droplets to rapidly move backward andforward in response to the electrical field (Figure 10.27). An oscillating droplet generates apressure wave with the same frequency as the alternating electric field which emanates fromit and can be detected by an ultrasonic transducer. The amplitude of the signal received bythe transducer is known as the electrokinetic sonic amplitude and is proportional to the


FIGURE 10.27 Principles of the electroacoustic technique.

dynamic mobility (µd) of the particles in the suspension. The dynamic mobility of theparticles is related to their zeta potential and size:

µε ε ξ

ηω ρ

ηdR G

r=

02

(10.17)

Here, εR is the relative permittivity of the continuous phase, ξ is the zeta potential, ω is theangular frequency, r is the droplet radius, and η and ρ are the viscosity and density of thecontinuous phase, respectively. G is a function that varies from 1 at low frequencies to 0 athigh frequencies. The frequency dependence of G is associated with the phase lag betweenthe alternating electrical field and the oscillating pressure wave generated by the particleswhich arises because of their inertia. At low frequencies, the particles move in-phase with theelectric field, and so the dynamic mobility is equal to the static mobility: µd = ε0εRξ/η ( i.e.,G = 1). At extremely high frequencies, the electric field alternates so quickly that the particleshave no time to respond and therefore remain stationary. Under these circumstances, noultrasonic pressure wave is generated by the particles (i.e., G = 0). The transition from thelow- to high-frequency regions occurs at a characteristic relaxation frequency (ωR) which isrelated to the size of the particles, decreasing as the size (inertia) of the particles increases.By measuring the frequency dependence of the dynamic mobility, it is therefore possible todetermine the droplet size. By measuring the dynamic mobility at low frequencies (where itis independent of droplet size), it is possible to determine the zeta potential.

Electroacoustic experiments can also be performed by applying an ultrasonic wave to anemulsion, which causes the droplets to oscillate backward and forward due to the densitydifference between themselves and the surrounding medium. An oscillating charged particlegenerates an alternating electric field that can be detected using suitable electrodes. Theamplitude of the signal received by the electrodes is called the colloid vibration potential, andit can also be used to determine the size and zeta potential of the emulsion droplets. The majoradvantage of electroacoustic techniques is that they can be applied to concentrated emulsions(φ ≤ 40%) without the need for any sample dilution. Nevertheless, they can only be used to


FIGURE 10.28 Dependence of the zeta potential of a 20% oil-in-water emulsion on the calciumconcentration measured using an electroacoustic technique. (Data from Colloidal Dynamics Ltd.,Application Note 203, Matec Applied Sciences, Hopkinton, MA.)

analyze emulsion droplets which are charged and have a significant density differencecompared to the surrounding liquid, and therefore they have limited application to many foodsystems.

The use of electroacoustics for measuring the zeta potential of droplets in concentratedemulsions is illustrated in Figure 10.28. When calcium ions are added to a 20% oil-in-wateremulsion stabilized by lecithin, they “bind” to the surface of the negatively charged dropletsand reduce the zeta potential. Above a certain concentration, they actually cause the sign ofthe droplet charge to become reversed. One would therefore expect low concentrations ofcalcium to reduce the electrostatic repulsion between droplets, which may lead to flocculation(Chapters 3 and 7).


References

Agboola, S.O. and Dalgleish, D.G., Calcium-induced destabilization of oil-in-water emulsions stabi-lized by caseinate or by β-lactoglobulin, Journal of Food Science, 60, 399, 1995.

Agboola, S.O. and Dalgleish, D.G., Kinetics of calcium-induced instability of oil-in-water emulsions:Studies under quiescent and shearing conditions, Lebensmittal Wissenschaft unt Technologie,29, 425 1996a.

Agboola, S.O. and Dalgleish, D.G., Enzymatic hydrolysis of milk proteins used for emulsion formation.1. Kinetics of protein breakdown and storage stability of the emulsions, Journal of Agricultureand Food Science, 44, 3631, 1996b.

Agboola, S.O. and Dalgleish, D.G., Enzymatic hydrolysis of milk proteins used for emulsion formation.2. Effects of calcium, pH and ethanol on the stability of the emulsions, Journal of Agricultureand Food Science, 44, 3637, 1996c.

Agboola, S.O. and Dalgleish, D.G., Effects of pH and ethanol on the kinetics of destabilization of oil-in-water emulsions containing milk proteins, Journal of the Science of Food and Agriculture,72, 448, 1996d.

Aguilera, J.M. and Kessler, H.G., Properties of mixed and filled-type dairy gels, Journal of FoodScience, 54, 1213, 1989.

Aguilera, J.M. and Stanley, D.W., Microstructural Principles of Food Processing and Engineering,Elsevier, Amsterdam, 1990.

Alaimo, M.H. and Kumosinski, T.F., Investigation of hydrophobic interactions in colloidal and biologi-cal systems by molecular dynamics simulations and NMR spectroscopy, Langmuir, 13, 2007,1997.

Alber, T., Stabilization energies of protein conformation, in Prediction of Protein Structure and thePrinciples of Protein Conformation, Fasman, G.D., Ed., Plenum Press, New York, 1989, 161.

Anandarajah, A. and Chen, J., Single correction function for computing retarded van der Waalsattraction, Journal of Colloid and Interface Science, 176, 293, 1995.

Aoki, H., Shirase, Y., Kato, J., and Watanabe, Y., Emulsion stabilizing properties of soy protein isolatesmixed with sodium caseinates, Journal of Food Science, 49, 212, 1984.

Arlauskas, R.A. and Weers, J.G., Sedimentation field-flow fractionation studies of composition ripen-ing in emulsion mixtures, Langmuir, 12, 1923, 1996.

Aronson, M.P., Surfactant induced flocculation of emulsions, in Emulsions — A Fundamental andPractical Approach, Sjoblom, J., Ed., Kluwer Academic Publishers, Netherlands, 1992, 75.

Asami, K., Evaluation of Colloids by Dielectric Spectroscopy, HP Application Note 380-3, HewlettPackard, Palo Alto, CA, 1995.

Atkins, P.W., Physical Chemistry, 5th ed., Oxford University Press, Oxford, 1994.Atkinson, P.J., Dickinson, E., Horne, D.S., and Richardson, R.M., Neutron reflectivity of adsorbed β-

casein and β-lactoglobulin at the air/water interface, Journal of the Chemical Society, FaradayTransactions, 91, 2847, 1995.

Aveyard, R., Binks, B.P., Clark, S., and Fletcher, P.D.I., Cloud points, solubilisation and interfacialtensions in systems containing nonionic surfactants, Chemical Technology and Biotechnology,48, 161, 1990.

Azzam, R.M. and Bachara, N.M., Ellipsometry and Polarized Light, North-Holland, Amsterdam, 1989.Baianu, I.C., Physical Chemistry of Food Processes, Vol. 1, Van Nostrand Reinhold, New York, 1992.Baines, Z.V. and Morris, E.R., Suppression of perceived flavor and taste by hydrocolloids, in Food

Colloids, Bee, R.D., Richmond, P., and Mingins, J., Eds., Royal Society of Chemistry, Cam-bridge, 1989, 184.

Baker, R., Controlled Release of Biologically Active Agents, John Wiley & Sons, New York, 1987.



Baker, E.N. and Hubbard, R.E., Hydrogen bonding in globular proteins, Progress in Biophysics andMolecular Biology, 44, 97, 1984.

Bakker, J., Flavor interactions with the food matrix and their effects on perception, in IngredientInteractions: Effects on Food Quality, Gaonkar, A.G., Ed., Marcel Dekker, New York, 1995.

Bakker, J. and Mela, D.J., Effect of emulsion structure on flavor release and taste perception, in Flavor–Food Interactions, McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Wash-ington, D.C., 1996, chap. 4.

Baldwin, R.L., Temperature dependence of the hydrophobic interaction in protein folding, Proceedingsof the National Academy of Sciences, 83, 8069, 1986.

Baldwin, R.W., Cloninger, M.R., and Lindsay, R.C., Flavor thresholds for fatty acids in bufferedsolutions, Journal of Food Science, 38, 528, 1973.

Banks, W. and Muir, D.D., Stability of alcohol containing emulsions, in Advances in Food Emulsionsand Foams, Dickinson, E. and Stainsby, G., Eds., Elsevier Applied Science, London, 1988,chap. 8.

Barford, N.M., Methods for characterization of structure in whippable dairy based emulsions, inCharacterization of Foods: Emerging Techniques, Gaonkar, A.G., Ed., Elsevier, Amsterdam,1995, chap. 3.

Barford, N.M. and Krog, M., Destabilization and fat crystallization of whippable emulsions (toppings)studied by pulsed NMR, Journal of the American Oil Chemists Society, 64, 112, 1987.

Barford, N.M., Krog, N., and Bucheim, W., Protein fat surfactant interactions in whippable emulsions,in Food Emulsions and Foams, Dickinson E., Ed., Royal Society of Chemistry, Cambridge,1987, 141.

Barford, N.M., Krog, N., Larsen, G., and Bucheim, W., Effects of emulsifiers on protein–fat interactionin ice cream mix during ageing. I. Quantitative analysis, Fat Science Technology, 93, 24, 1991.

Barnes, H.A., Rheology of emulsions — A review, Colloids and Surfaces, 91, 89, 1994.Barylko-Pikielna, N., Martin, A., and Mela, D.J., Perception of taste and viscosity of oil-in-water and

water-in-oil emulsions, Journal of Food Science, 59, 1318, 1994.Basaran, T., Demetriades, K., and McClements, D.J., Ultrasonic imaging of gravitational separation in

emulsions, Colloids and Surfaces, submitted, 1998.Becher, P., Emulsions: Theory and Practice, Reinhold Publishing, New York, 1957.Becher, P., Encyclopedia of Emulsion Technology, Vol. 1, Marcel Dekker, New York, 1983.Becher, P., Hydrophile–lipophile balance: An updated bibliography, in Encyclopedia of Emulsion

Technology, Vol. 2, Becher, P., Ed., Marcel Dekker, New York, 1985.Becher, P., HLB: Update III, in Encyclopedia of Emulsion Technology, Vol. 4, Becher, P., Ed., Marcel

Dekker, New York, 1996.Bell, G.A., Molecular mechanisms of olfactory perception: Their potential for future technologies,

Trends in Food Science and Technology, 7, 425, 1996.BeMiller, J.N. and Whistler, R.L., Carbohydrates, in Food Chemistry, 3rd ed., Fennema, O.R., Ed.,

Marcel Dekker, New York, 1996, 157.Ben-Naim, A., Hydrophobic Interactions, Plenum Press, New York, 1980.Bentley, B.J. and Leal, L.G., An experimental investigation of drop deformation and breakup in steady,

two-dimensional linear flows, Journal of Fluid Mechanics, 167, 241, 1986.Bergenstahl, G., Physiochemical aspects of an emulsifier functionality, in Food Emulsifiers and Their

Applications, Hasenhuettl, G.L. and Hartel, R.W., Eds., Chapman and Hall, New York, 1997,chap. 6.

Berger, K.G., Ice cream, in Food Emulsions, Friberg, S., Ed., Marcel Dekker, New York, 1976, chap.4.

Bergethon, P.R. and Simons, E.R., Biophysical Chemistry: From Molecules to Membranes, Springer-Verlag, New York, 1990.

Besseling, N.A.M., Theory of hydration forces between surfaces, Langmuir, 13, 2113, 1997.Bibette, J., Depletion interactions and fractionated crystallization for polydisperse emulsion purifica-

tion, Journal of Colloid and Interface Science, 147, 474, 1991.Bijsterbosch, B.H., Bos, M.T.A., Dickinson, E., van Opheusden, J.H.J., and Walstra, P., Brownian

dynamics simulation of particle gel formation: From argon to yoghurt, Faraday Discussions,101, 51, 1995.


Binks, B.P., Emulsion type below and above the CMC in AOT microemulsion systems, Colloids andSurfaces, 71, 167, 1993.

Birch, G.G. and Lindley, M.G., Interactions of Food Components, Elsevier Applied Science, London,1986.

Birker, P.J.M.W.L. and Padley F.B., Physical properties of fats and oils, in Recent Advances inChemistry and Technology of Fats and Oils, Hamilton, R.J. and Bhati, A., Eds., ElsevierApplied Science, London, 1987, chap. 1.

Blonk, J.C.G. and van Aalst, H., Confocal scanning light microscopy in food research, Food ResearchInternational, 26, 297, 1993.

Boelens, M.H. and van Gemert, L.J., Physicochemical parameters related to organoleptic properties offlavour components, in Developments in Food Flavours, Birch, G.G. and Lindley, M.G., Eds.,Elsevier Applied Science, London, 1986, 23.

Bohren, C.F. and Huffman, D.R., Adsorption and Scattering of Light by Small Particles, John Wiley& Sons, New York, 1983.

Boistelle, R., Fundamentals of nucleation and crystal growth, in Crystallization and Polymorphism ofFats and Fatty Acids, Garti, N. and Sato, K., Eds., Marcel Dekker, New York, 1988, chap. 5.

Bomben, J.L., Bruin, S., Thijssen, H.A.C., and Merson, R.L., Aroma recovery and retention in concen-tration and drying of foods, Advances in Food Research, 20, 1, 1973.

Boode, K., Partial Coalescence in Oil-in-Water Emulsions, Ph.D. thesis, Wageningen AgriculturalUniversity, Wageningen, Netherlands, 1992.

Boode, K. and Walstra, P., Partial coalescence in oil-in-water emulsions. 1. Nature of the aggregation,Colloids and Surfaces A, 81, 121, 1993a.

Boode, K. and Walstra, P., The kinetics of partial coalescence in oil-in-water emulsions, in FoodColloids and Polymers: Stability and Mechanical Properties, Dickinson, E. and Walstra, P.,Eds., Royal Society of Chemistry, Cambridge, 1993b, 23.

Boode, K., Bispernick, C., and Walstra, P., Destabilization of O/W emulsions containing fat crystals bytemperature cycling, Colloids and Surfaces, 61, 55, 1991.

Boode, K., Walstra, P., and Degrootmostert, A.E.A., Partial coalescence in oil-in-water emulsions. 1.Influence of the properties of the fat, Colloids and Surfaces A, 81, 139, 1993.

Bos, M., Nylander, T., Arnebrant, T., and Clark, D.C., Protein/emulsifier interactions, in Food Emul-sifiers and Their Applications, Hasenhuettl, G.L. and Hartel, R.W., Eds., Chapman and Hall,New York, 1997, chap. 5.

Boudaud, N. and Dumont, J.P., Interaction between flavor components and β-lactoglobulin, in Flavor–Food Interactions, McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Wash-ington, D.C., 1996, chap. 4.

Bourne, M.C., Food Texture and Viscosity: Concept and Measurement, Academic Press, New York,1982.

Bower, C., Washington, C., and Purewal, T.S., The use of image analysis to characterize aggregates ina shear field, Colloids and Surfaces A, 127, 105, 1997.

Bradley, R.L., Moisture and total solids analysis, in Introduction to the Chemical Analysis of Foods,Nielsen, S.S., Ed., Bartlett and Jones, Boston, 1994, chap. 7.

Brady, J.W., The role of dynamics and solvation in protein structure and function, in Food Proteins,Kinsella, J.E. and Soucie, W.G., Eds., American Oil Chemists Society, Champaign, IL, 1989,chap. 1.

Brady, J.W. and Ha, S.N., Molecular dynamic simulations of the aqueous solvation of sugars, in WaterRelationships of Foods, Duckworth, R.B., Ed., Academic Press, London, 1975, 3.

Braginsky, L.M. and Belevitskaya, M.A., Kinetics of droplets breakup in agitated vessels, in Liquid–Liquid Systems, Kulov, N.N., Ed., Nova Science, Commack, NY, 1996, chap. 1,

Brakenhoff, G.J., van der Voort, H.T.M., van Spronsen, E.A., and Nanninga, N., Three-dimensionalimaging of biological structures by high resolution confocal laser scanning microscopy, Scan-ning Microscopy, 2, 33, 1988.

Brash, J.L. and Horbett, T.A., Proteins at Interfaces, American Chemical Society, Washington, D.C.,1987.

Bremer, L.G.B., Fractal Aggregation in Relation to Formation and Properties of Particle Gels, Ph.D.thesis, Wageningen Agricultural University, Wageningen, Netherlands, 1992.



Bremer, L.G.B., Bijsterbosch, B.H., Walstra, P., and Van Vliet, T., Formation, properties and fractalstructure of particle gels, Advances in Colloid and Interface Science, 46, 117, 1993.

Brennan, J.G., Butters, J.R., Cowell, N.D., and Lilly, A.E.V., Food Engineering Operations, AppliedScience Publishers, London, 1981.

Britten, M. and Giroux, H.J., Coalescence index of protein-stabilized emulsions, Journal of FoodScience, 56, 792, 1991.

Brooker, B.E., Imaging food systems by confocal laser scanning microscopy, in New PhysicochemicalTechniques for the Characterization of Complex Food Systems, Dickinson, E., Ed., BlackieAcademic and Professional, London, 1995, chap. 2.

Brossard, C., Rousseau, F., and Dumont, J.P., Flavour release and flavour perception in oil-in-wateremulsions: Is the link so close? in Flavour Science: Recent Developments, Tayloer, A.J. andMottram, D.S., Eds., Royal Society of Chemistry, Cambridge, 1996, 375.

Bucheim, W. and Dejmek, P., Milk and dairy-type emulsions, in Food Emulsions, 2nd ed., Larsson, K.and Friberg, S., Eds., Marcel Dekker, New York, 1990, chap. 6.

Buffler, C.R., Advances in dielectric characterization of foods, in Characterization of Food: EmergingMethods, Gaonkar, A.G., Ed., Elsevier, Amsterdam, 1995, chap. 10.

Bujannunez, M.C. and Dickinson, E., Brownian dynamics simulation of a multisubunit deformableparticle in simple shear-flow, Journal of the Chemical Society, Faraday Transactions, 90,2737, 1994.

Buttery, R.G., Ling, L.C., and Guadagni, D.G., Volatilities of aldehydes, ketones and esters in dilutewater solution, Journal of Agricultural and Food Chemistry, 17, 385, 1969.

Buttery, R.G., Bomden, J.L., Guadagni, D.G., and Ling, L.C., Some considerations of the volatilitiesof organic flavor compounds in foods, Journal of Agricultural and Food Chemistry, 19, 1045,1971.

Buttery, R.G., Guadagni, D.G., and Ling, L.C., Flavor compounds: Volatilities in vegetable oil and oil–water mixtures. Estimation of odor thresholds, Journal of Agricultural and Food Chemistry, 21,198, 1973.

Callaghan, P.T., Jolley, K.W., and Humphrey, R.S., Diffusion of fat and water in cheese as studied bypulsed field gradient nuclear magnetic resonance, Journal of Colloid and Interface Science, 93,521, 1983.

Campanella, O.H., Dorward, N.M., and Singh, H., A study of the rheological properties of concentratedfood emulsions, Journal of Food Engineering, 25, 427, 1995.

Campbell, I.J., The role of fat crystals in emulsion stability, in Food Colloids, Bee, R.D., Richmond,P., and Mingins, J., Eds., Royal Society of Chemistry, Cambridge, 1991, 272.

Campbell, I., Norton, I., and Morley, W., Factors controlling the phase inversion of oil-in-wateremulsions, Netherlands Milk and Dairy Journal, 50, 167, 1996.

Carasso, M.L., Rowlands, W.N., and Kennedy, R.A., Electroacoustic determination of droplet size andzeta potential in concentrated intravenous fat emulsion, Journal of Colloid and InterfaceScience, 174, 405, 1995.

Carnie, S.L., Chan, D.Y.C., and Stankovich, J., Computation of forces between spherical colloidalparticles: Non-linear Poisson–Boltzmann theory, Journal of Colloid and Interface Science, 165,116, 1994.

Carr, J.M., Sufferling, K., and Poppe, J., Hydrocolloids and their use in the confectionery industry, FoodTechnology, 49(7), 41, 1995.

Carr, J., Baloga, D., Guinard, J.-X., Lawter, L., Marty, C., and Squire, C., The effect of gelling agenttype and concentration on flavor release in model systems, in Flavor–Food Interactions,McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Washington, D.C., 1996,chap. 9.

Cebula, D.J., McClements, D.J., Povey, M.J.W., and Smith, P.R., Neutron diffraction studies of liquidand crystalline trilaurin, Journal of the American Oil Chemists Society, 69, 130, 1992.

Cesero, A., The role of conformation on the thermodynamics and rheology of aqueous solutions ofcarbohydrate polymers, in Water in Foods: Fundamental Aspects and Their Significance inRelation to the Processing of Foods, Fito, P., Mullet, A., and McKenna, B., Eds., ElsevierApplied Science, London, 1994, 27.

Chang, C.M., Powries, W.D., and Fennema, O., Electron microscopy of mayonnaise, Canadian Instituteof Food Science and Technology Journal, 5, 134, 1972.


Chapman, D., The Structure of Lipids, Methuen, London, 1965.Charalambous, G. and Doxastakis, G., Food Emulsifiers: Chemistry, Technology, Functional Proper-

ties and Applications, Elsevier, Amsterdam, 1989.Chen, J. and Anadarajah, A., Van der Waals attraction between spherical particles, Journal of Colloid

and Interface Science, 180, 519, 1996.Chen, J. and Dickinson, E., Time-dependent competitive adsorption of milk proteins and surfactants in

oil-in-water emulsions, Journal of the Science of Food and Agriculture, 62, 283, 1993.Chen, J., Dickinson, E., and Iveson, G., Interfacial interactions, competitive adsorption and emulsion

stability, Food Microstructure, 12, 135, 1993.Chow, C.K., Fatty Acids in Foods and Their Health Implication, Marcel Dekker, New York, 1992.Christenson, H.K., Fang, J., Ninham, B.W., and Parker, J.L., Effect of divalent electrolyte on the

hydrophobic attraction, Journal of Physical Chemistry, 94, 8004, 1990.Christian, S.D. and Scamehorn, J.F., Solubilization in Surfactant Aggregates, Marcel Dekker, New

York, 1995.Chrysam, M.M., Margarine and spreads, in Bailey’s Industrial Oil and Fat Products, Vol. 3, 5th ed.,

Hui, Y.H., Ed., John Wiley, New York, 1996, chap. 2.Claesson, P.M., Experimental evidence for repulsive and attractive forces not accounted for by conven-

tional DLVO theory, Progress in Colloid and Polymer Science, 74, 48, 1987.Claesson, P.M. and Christenson, H.K., Very long range attractive forces between uncharged hydrocar-

bon and fluorocarbon surfaces in water, Journal of Physical Chemistry, 92, 1650, 1988.Claesson, P.J., Blombert, E., Froberg, J.C., Nylander, T., and Arnebrant, T., Protein interactions at solid

surfaces, Advances in Colloid and Interface Science, 57, 161, 1995.Claesson, P.M., Ederth, T., Bergeron, V., and Rutland, M.W., Techniques for measuring surface forces,

Advances in Colloid and Interface Science, 67, 119, 1996.Clark, A.H., The application of network theory to food systems, in Food Structure and Behaviour,

Blanshard, J.M.V. and Lillford, P.J., Eds., Academic Press, London, 1987, chap. 2.Clark, A.H. and Lee-Tuffnell, C.D., Gelation of globular proteins, in Functional Properties of Food

Macromolecules, Mitchell, J.R. and Ledward, D.A., Eds., Elsevier Applied Science, London,1986, chap. 5.

Clausse, D., Dielectric properties of emulsions and related systems, in Encyclopedia of EmulsionTechnology, Vol. 1, Blecher, P., Ed., Marcel Dekker, New York, 1983, chap. 9.

Clause, D., Research techniques utilizing emulsions, in Encyclopedia of Emulsion Technology, Vol. 2,Becher, P., Ed., Marcel Dekker, New York, 1985, chap. 2.

Clydesdale, F.M., Methods and measurements of food color, in Theory, Determination and Control ofPhysical Properties of Food Materials, Rha, C., Ed., D. Reidel Publishing, Boston, 1975, chap.14.

Clydesdale, F.M. and Francis, F.J., Food Colorimetry: Theory and Applications, AVI Publishing,Westport, CT, 1975.

Corredig, M. and Dalgleish, D.G., A differential microcalorimetric study of whey proteins and theirbehaviour in oil-in-water emulsions, Colloids and Surfaces B, 4, 411, 1995.

Coultate, T.P., Food: The Chemistry of Its Components, 3rd ed., Royal Society of Chemistry, Cam-bridge, 1996.

Couper, A., Surface tension and its measurement, in Physical Methods of Chemistry, Vol. IXA,Rossiter, B.W. and Baetzold, R.C., Eds., John Wiley & Sons, New York, 1993, chap. 1.

Coupland, J.N. and McClements, D.J., Lipid oxidation in food emulsions, Trends in Food Science andTechnology, 7, 83, 1996.

Coupland, J.N. and McClements, D.J., Droplet size determination in food emulsions: Comparison ofultrasonic and light scattering techniques, Journal of Food Engineering, submitted, 1998.

Coupland, J.N., Dickinson, E., McClements, D.J., Povey, M.J.W, and Rancourt de Mimmerand, C.,Crystallization of simple paraffins and monoacid saturated triacylglycerols dispersed as an oilphase in water, in Food Colloids and Polymers: Structure and Dynamics, Dickinson, E. andWalstra, P., Eds., Royal Society of Chemistry, Cambridge, 1993.

Coupland, J.N., Zhu, Z., Wan, H., McClements, D.J., Nawar, W.W., and Chinachotti, P., Dropletcomposition affects the rate of oxidation of emulsified ethyl linoleate, Journal of the AmericanOil Chemists Society, 73, 795, 1996.



Coupland, J.N., Brathwaite, D., Fairley, P., and McClements, D.J., Effect of ethanol on the solubiliza-tion of hydrocarbon emulsion droplets in nonionic surfactant micelles, Journal of Colloid andInterface Science, 190, 71, 1997.

Courthaudon, J.-L., Dickinson, E., and Dalgleish, D.G., Competitive adsorption of β-casein and non-ionic surfactants in oil-in-water emulsions, Journal of Colloid and Interface Science, 145, 390,1991a.

Courthaudon, J.-L., Dickinson, E., Matsumura, Y., and Clark, D.C., Competitive adsorption of β-lactoglobulin and polyoxyethylene sorbitan monostearate 20 at the oil–water interface, Colloidsand Surfaces, 56, 293, 1991b.

Courthaudon, J.-L., Dickinson, E., Matsumura, Y., and Williams, A., Influence of emulsifier on thecompetitive adsorption of whey proteins in emulsions, Food Structure, 10, 109, 1991c.

Courthaudon, J.-L., Dickinson, E., and Christie, W.W., Competitive adsorption of lecithin and β-caseinin oil-in-water emulsions, Journal of Agriculture and Food Chemistry, 39, 1365, 1991d.

Cramer, C.J. and Tuhlar, D.G., Structure and Reactivity in Aqueous Solution: Characterization ofChemical and Biological Systems, American Chemical Society, Washington, D.C., 1994.

Crank, J., The Mathematics of Diffusion, Oxford University Press, London, 1975.Curle, N. and Davies, H.J., Modern Fluid Dynamics, Vol. 1, D van Nostrand, London, 1968.Dalgleish, D.G., Protein stabilized emulsions and their properties, in Water and Food Quality, Hardman,

T.M., Ed., Elsevier Applied Science, London, 1989, chap. 6.Dalgleish, D.G., Structure and properties of adsorbed layers in emulsions containing milk proteins, in

Food Macromolecules and Colloids, Dickinson, E. and Lorient, D., Eds., Royal Society ofChemistry, Cambridge, 1995, 23.

Dalgleish, D.G., Food emulsions, in Emulsions and Emulsion Stability, Sjoblom, J., Ed., MarcelDekker, New York, 1996a.

Dalgleish, D.G., Conformations and structures of milk proteins adsorbed to oil–water interfaces, FoodResearch International, 29, 541, 1996b.

Dalgleish, D.G. and Hallet, F.R., Dynamic light scattering — Applications to food systems, FoodResearch International, 28, 181, 1995.

Damodaran, S., Interrelationship of molecular and functional properties of food proteins, in FoodProteins, Kinsella, J.E. and Soucie, W.G., Eds., American Chemical Society, Champaign, IL,1989, chap. 3.

Damodaran, S., Interfaces, protein films and foams, in Advances in Food and Nutrition Research, Vol.34, Kinsella, J.E., Ed., Academic Press, San Diego, 1990, 1.

Damodaran, S., Structure–function relationships of food proteins, in Protein Functionality in FoodSystems, Hettiarchchy, N. and Ziegler, G.R., Eds., Marcel Dekker, New York, 1994, 1.

Damodaran, S., Amino acids, peptides and proteins, in Food Chemistry, 3rd ed., Fennema, O.R., Ed.,Marcel Dekker, New York, 1996, 321.

Damodaran, S., Food proteins: An overview, in Food Proteins and Their Applications, Marcel Dekker,New York, 1997, chap. 1.

Darling, D.F., Recent advances in the destabilization of dairy emulsions, Journal of Dairy Research,49, 695, 1982.

Das, A.K. and Ghosh, P.K., Concentrated emulsions — Investigation of polydispersity and dropletdistortion and their effect on volume fraction and interfacial area, Langmuir, 6, 1668, 1990.

Das, K.P. and Kinsella, J.E., Stability of food emulsions: Physicochemical role of protein and non-protein emulsifiers, Advances in Food and Nutrition Research, 34, 81, 1990.

Davis, E.A., Thermal analysis, in Introduction to the Chemical Analysis of Foods, Nielsen, S.S., Ed.,Bartlett and Jones, Boston, 1994a, chap. 34.

Davis, H.T., Factors determining emulsion type: Hydrophile–lipophile balance and beyond, Colloidsand Surfaces A, 91, 9, 1994b.

Davis, R.H., Velocities of sedimenting particles in suspensions, in Sedimentation of Small Particles ina Viscous Liquid, Tory, E.M., Ed., Computational Mechanics Publications, Southampton, U.K.,1996, chap. 6.

Davis, R.H., Schonberg, J.A., and Rallison, J.M., The lubrication force between two viscous drops,Physics of Fluids A, 1, 77, 1989.

Dea, I.C.M., Polysaccharide conformation in solutions and gels, in Food Carbohydrates, Lineback,D.R. and Inglett, G.E., Eds., AVI Publishing, Westport, CT, 1982, chap. 22.


Deffenbaugh, L.B., Carbohydrate/emulsifier interactions, in Food Emulsifiers and Their Applications,Hasenhuettl, G.L. and Hartel, R.W., Eds., Chapman and Hall, New York, 1997, chap. 4.

Demetriades, K., Coupland, J.N., and McClements, D.J., Physical properties of whey protein stabilizedemulsions as related to pH and NaCl, Journal of Food Science, 62, 342, 1997a.

Demetriades, K., Coupland, J.N., and McClements, D.J., Physicochemical properties of whey proteinstabilized emulsions as affected by heating and ionic strength, Journal of Food Science, 62,462, 1997b.

Derjaguin, B.V., Theory of Stability of Colloids and Thin Films, Consultants Bureau, New York, 1989.Derjaguin, B.V., Churaev, N.V., and Muller, V.M., Surface Forces, Consultants Bureau, New York,

1987.de Roos, A.L. and Walstra, P., Loss of enzyme activity due to adsorption onto emulsion droplets,

Colloids and Surfaces B, 6, 201, 1996.Desrosier, N.W., Elements of Food Technology, AVI Publishing, Westport, CT, 1977, chap. 13.de Vries, A.J., Effect of particle aggregation on the rheological behaviour of disperse systems, in

Rheology of Emulsions, Sherman, P., Ed., Pergamon Press, New York, 1963.de Wit, J.N. and van Kessel, T., Effects of ionic strength on the solubility of whey protein products.

A colloid chemical approach, Food Hydrocolloids, 10, 143, 1996.Dickinson, E., A model of a concentrated dispersion exhibiting bridging flocculation and depletion

flocculation, Journal of Colloid and Interface Science, 132, 274, 1989.Dickinson, E., Introduction to Food Colloids, Oxford University Press, Oxford, 1992.Dickinson, E., Protein–polysaccharide interactions in food colloids, in Food Colloids and Polymers:

Stability and Mechanical Properties, Dickinson, E. and Walstra, P., Royal Society of Chem-istry, Cambridge, 1993, 77.

Dickinson, E., Colloidal aspects of beverages, Food Chemistry, 51, 343, 1994.Dickinson, E., New Physico-Chemical Techniques for the Characterization of Complex Food Systems,

Chapman & Hall, London, 1995a.Dickinson, E., Recent trends in food colloids research, in Food Macromolecules and Colloids, Dickinson,

E. and Lorient, D., Eds., Royal Society of Chemistry, Cambridge, 1995b, 19.Dickinson, E. and Euston, S.R., Computer-simulation of bridging flocculation, Journal of the Chemical

Society, Faraday Transactions, 87, 2193, 1991.Dickinson, E. and Golding, M., Rheology of sodium caseinate stabilized oil-in-water emulsions, Jour-

nal of Colloid and Interface Science, 191, 166, 1997a.Dickinson, E. and Golding, M., Depletion flocculation of emulsions containing unadsorbed sodium

caseinate, Food Hydrocolloids, 11, 13, 1997b.Dickinson, E. and Hong, S.-K., Surface coverage of β-lactoglobulin at the oil–water interface: Influence

of protein heat treatment and various emulsifiers, Journal of Agricultural and Food Chemistry,42, 1602, 1994.

Dickinson, E. and Hong, S.T., Interfacial and stability properties of emulsions: Influence of protein heattreatment and emulsifiers, in Food Macromolecules and Colloids, Dickinson, E. and Lorient,D., Eds., Royal Society of Chemistry, Cambridge, 1995a, 269.

Dickinson, E. and Hong, S.T., Influence of water-soluble nonionic emulsifier on the rheology of heat-setprotein-stabilized emulsion gels, Journal of Agricultural and Food Chemistry, 43, 2560, 1995b.

Dickinson, E. and Iveson, G., Adsorbed films of β-lactoglobulin + lecithin at the hydrocarbon–waterand triglyceride–water interfaces, Food Hydrocolloids, 6, 553, 1993.

Dickinson, E. and Matsumura, Y., Time-dependent polymerization of β-lactoglobulin through disulphidebonds at the oil–water interface in emulsions, International Journal of Biology and Macromol-ecules, 13, 26, 1991.

Dickinson, E. and McClements, D.J., Advances in Food Colloids, Chapman & Hall, London, 1995.Dickinson, E. and Stainsby, G., Colloids in Foods, Applied Science Publishers, London, 1982.Dickinson, E. and Stainsby, G, Emulsion stability, in Advances in Food Emulsions and Foams, Dickinson,

E. and Stainsby, G., Eds., Elsevier Applied Science, London, 1988, chap. 1.Dickinson, E. and Tanai, S., Temperature dependence of the competitive displacement of protein from

the emulsion droplet surface by surfactants, Food Hydrocolloids, 6, 163, 1992.Dickinson, E. and Tong, S.-T., Interfacial and stability properties of emulsions: Influence of protein heat

treatment and emulsifiers, in Food Macromolecules and Colloids, Dickinson, E. and Lorient,D., Eds., Royal Society of Chemistry, Cambridge, 1995, 269.



Dickinson, E. and Williams, A., Orthokinetic coalescence of protein-stabilized emulsions, Colloids andSurfaces A, 88, 314, 1994.

Dickinson, E. and Yamamoto Y., Viscoelastic properties of heat-set whey protein stabilized emulsiongels with added lecithin, Journal of Food Science, 61, 811, 1996.

Dickinson, E., Murray, B.S., and Stainsby, G., Coalescence stability of emulsion sized droplets at aplanar oil–water interface and the relationship to protein film surface rheology, Journal of theChemical Society, Faraday Transactions 1, 84, 871, 1988.

Dickinson, E., Goller, M.I., McClements, D.J., and Povey, M.J.W., Ultrasonic monitoring of crystal-lization in oil-in-water emulsions, Journal of the Chemical Society, Faraday Transactions, 86,1147, 1990.

Dickinson, E., McClements, D.J., and Povey, M.J.W., Ultrasonic investigation of the particle sizedependence of crystallization in n-hexadecane-in-water emulsions, Journal of Colloid andInterface Science, 142, 103, 1991.

Dickinson, E., Hunt, J.A., and Horne, D.S., Calcium induced flocculation of emulsions containingadsorbed β-casein or phosvitin, Food Hydrocolloids, 6, 359, 1992.

Dickinson, E., Owusu, R.K., Tan, S., and Williams, A., Oil-soluble surfactants have little effect oncompetitive adsorption of α-lactalbumin and β-lactoglobulin in emulsions, Journal of FoodScience, 58, 295, 1993a.

Dickinson, E., Owusu, R.K., and Williams, A., Orthokinetic destabilization of a protein stabilizedemulsion by a water-soluble surfactant, Journal of the Chemical Society, Faraday Transac-tions, 89, 865, 1993b.

Dickinson, E., Iveson, G., and Tanai, S., Competitive adsorption in protein stabilized emulsions con-taining oil-soluble and water-soluble surfactants, in Food Colloids and Polymers: Stability andMechanical Properties, Dickinson, E. and Walstra, P., Eds., Royal Society of Chemistry,Cambridge, 1993c, 312.

Dickinson, E., Goller, M.I., and Wedlock, D.J., Creaming and rheology of emulsions containingpolysaccharide and non-ionic or anionic surfactants, Colloids and Surfaces A, 75, 195, 1993d.

Dickinson, E., Goller, M.I., and Wedlock, D.J., Osmotic pressure, creaming and rheology of emulsionscontaining non-ionic polysaccharide, Journal of Colloid and Interface Science, 172, 192, 1995.

Dickinson, E., Hong, S.T., and Yamanoto, Y., Rheology of heat-set emulsion gels containing β-lactoglobulin and small-molecule surfactants, Netherlands Milk and Dairy Journal, 50, 199,1996.

Dickinson, E., Golding, M., and Povey, M.J.W., Rheology and dynamics of water-in-oil emulsionscontaining sodium caseinate, Journal of Colloid and Interface Science, 185, 515, 1997.

Dill, K.A., Dominant forces in protein folding, Biochemistry, 29, 7133, 1990.Doi, E., Gels and gelling of globular proteins, Trends in Food Science and Technology, 4, 1, 1993.Ducker, W.A. and Pashley, R.M., Forces between mica surfaces in the presence of rod-shaped divalent

counterions, Langmuir, 8, 109, 1992.Duckworth, R.B., Water Relationships of Foods, Academic Press, London, 1975.Dukhin, A.S. and Goetz, P.J., Acoustic and electroacoustic spectroscopy, Langmuir, 12, 4336, 1996.Dukhin, S. and Sjoblom, J., Kinetics of Brownian and gravitational coagulation in dilute emulsions, in

Emulsions and Emulsion Stability, Sjoblom, J., Ed., Marcel Dekker, New York, 1996.Dukhin, S.S., Krietchamer, G., and Miller R., Dynamics of Adsorption at Liquid Interfaces: Theory,

Experiment and Application, Elsevier, Amsterdam, 1995.Eads, T.M., Molecular origins of structure and functionality in foods, Trends in Food Science and

Technology, 5, 147, 1994.Eagland, D., Nucleic acids, peptides and proteins, in Water: A Comprehensive Treatise, Vol. 4, Franks,

F., Ed., Plenum Press, New York, 1975, chap. 5.Eastoe, J., Small-angle neutron scattering and neutron reflection, in New Physicochemical Techniques

for the Characterization of Complex Food Systems, Dickinson, E., Ed., Blackie Academic andProfessional, London, 1995, chap. 12.

Edwards, S.F., Lillford, P.J., and Blanshard, J.M.V., Gels and networks in practice and theory, in FoodStructure and Behaviour, Blanshard, J.M.V. and Lillford, P.J., Eds., Academic Press, London,1987, chap. 1.

Eisenberg, D. and Kauzmann, W., The Structure and Properties of Water, Oxford University Press,Oxford, 1969.


Elworthy, P.H., Florence, A.T., and Macfarlane, C.B., Solubilization by Surface-Active Agents and ItsApplication in Chemistry and the Biological Sciences, Chapman and Hall, London, 1968.

Epstein, P.S. and Carhart, R.R., The absorption of sound in suspensions and emulsions, Journal of theAcoustical Society of America, 25, 553, 1953.

Esselink, K., Hilbers, P.A.J., van Os, N.M., Smit, B., and Karaborni, S., Molecular dynamics simula-tions of mode oil/water/surfactant systems, Colloids and Surfaces A, 91, 155, 1994.

Evans, D.F. and Wennerstrom, H., The Colloidal Domain: Where Physics, Chemistry, Biology andTechnology Meet, VCH Publishers, New York, 1994.

Everett, D.H., Basic Principles of Colloid Science, Royal Society of Chemistry, Cambridge, 1988.Evison, J., Dickinson, E., Owusu Apenten, R.K., and Williams, A., Formulation and properties of

protein stabilized water-in-oil-in-water multiple emulsions, in Food Macromolecules andColloids, Dickinson, E. and Lorient, D., Eds., Royal Society of Chemistry, Cambridge, 1995,235.

Farinato, R.S. and Rowell, R.L., Optical properties of emulsions, in Encyclopedia of Emulsion Tech-nology, Vol. 1, Becher, P., Marcel Dekker, New York, 1983, chap. 8.

Fellows, P., Food Processing Technology: Principles and Practice, VCH Publishers, Weinheim, Ger-many, 1988, chap. 3 and 4.

Fennema, O.R., Water and protein hydration, in Food Proteins, AVI Publishing, Westport, CT, 1977,chap. 3.

Fennema, O.R., Ed., Food Chemistry, 3rd ed., Marcel Dekker, New York, 1996a.Fennema, O.R., in Food Chemistry, 3rd ed., Fennema, O.R., Ed., Marcel Dekker, New York, 1996b,

chap. 2.Fennema, O.R. and Tannenbaum, S.R., Introduction to food chemistry, in Food Chemistry, 3rd ed.,

Fennema, O.R., Ed., Marcel Dekker, New York 1996, 1.Fisher, L.R. and Parker, N.S., How do food emulsion stabilizers work? CSIRO Food Research Quar-

terly, 45, 33, 1985.Fito, P., Mullet, A., and McKenna, B., Water in Foods: Fundamental Aspects and Their Significance

in Relation to the Processing of Foods, Elsevier Applied Science, London, 1994.Formo, M.W., Physical properties of fats and fatty acids, in Bailey’s Industrial Oil and Fat Products,

Vol. 1, 5th ed., Swern, D., Ed., John Wiley & Sons, New York, 1979, chap. 3.Fox, P.F. and Condon, J.J., Food Proteins, Applied Science Publishers, London, 1982.Francis, F.J., Basic concepts of colorimetry, in Theory, Determination and Control of Physical Prop-

erties of Food Materials, Rha, C., Ed., D. Reidel Publishing, Boston, 1975, chap. 14.Francis, F.J., Colorimetric properties of foods, in Engineering Properties of Foods, 2nd ed., Rao, M.A.

and Rizvi, S.S.H., Eds., Marcel Dekker, New York, 1995, chap. 10.Francis, F.J. and Clydesdale, F.M., Food Colorimetry: Theory and Applications, AVI Publishing,

Westport, CT, 1975.Frankel, E.N., Recent advances in lipid oxidation, Journal of Food Science and Agriculture, 54, 495,

1991.Frankel, E.N., Evaluation of antioxidant activity of rosemary extracts, carnosol and carnosic acid in bulk

vegetable oils and fish oil and their emulsions, Journal of the Science of Food and Agriculture,72, 201, 1996.

Frankel, E.N., Huang, S.W., Kanner, J., and German, J.B., Interfacial phenomena in the evaluation ofantioxidants: Bulk oils versus emulsions, Agriculture and Food Chemistry, 42, 1054, 1994.

Franklin, F., College Physics, 4th ed., Harcourt Brace Jovanovich, New York, 1977.Franks, F., Water: A Comprehensive Treatise, Vol. 1–7, Plenum Press, New York, 1972–82.Franks, F., The solvent properties of water, in Water: A Comprehensive Treatise, Vol. 2, Franks, F., Ed.,

Plenum Press, New York, 1973, chap. 1.Franks, F., The hydrophobic interaction, in Water: A Comprehensive Treatise, Vol. 4, Franks, F., Ed.,

Plenum Press, New York, 1975a, chap. 1.Franks, F., Water, ice and solutions of simple molecules, in Water Relationships of Foods, Duckworth,

R.B., Ed., Academic Press, London, 1975b, 3.Franks, F., Water and aqueous solutions: Recent advances, in Properties of Water in Foods in Relation

to Quality and Stability, Simatos, D. and Multon, J.L., Eds., Martinus Nijhoff Publishers,Dordrecht, 1985a, 1.



Franks, F., Complex aqueous systems at subzero temperatures, in Properties of Water in Foods inRelation to Quality and Stability, Simatos, D. and Multon, J.L., Eds., Martinus Nijhoff Publish-ers, Dordrecht, 1985b, 497.

Franks, F., Hydration phenomena: An update and implications for the food processing industry, inWater Relationships in Foods: Advances in the 1980’s and Trends for the 1990’s, Levine, H.and Slade, L., Eds., Plenum Press, New York, 1991, 1.

Franzen, K.L. and Kinsella, J.E., Physicochemical aspects of food flavouring, Chemistry and Industry,21, 505, 1975.

Friberg, S., Emulsion stability, in Food Emulsions, 3rd ed., Friberg, S. and Larsson, K., Eds., MarcelDekker, New York, 1997, chap. 1.

Friberg, S.E. and El-Nokaly, M., Multilayer emulsions, in Physical Properties of Foods, Peleg, M. andBagley, E.B., Eds., AVI Publishing, Westport, CT, 1983, chap. 5.

Friberg, S. and Larsson, K., Eds., Food Emulsions, 3rd ed., Marcel Dekker, New York, 1997.Friedman, M., The Chemistry and Biochemistry of the Sulfhydryl Group in Amino Acids, Peptides and

Proteins, Pergamon Press, Oxford, 1973.Galema, S.A. and Hoiland, H., Stereochemical aspects of hydration of carbohydrates in aqueous

solutions: Density and ultrasonic measurements, Journal of Physical Chemistry, 95, 5321,1991.

Gaonkar, A., Characterization of Foods: Emerging Methods, Elsevier, Amsterdam, 1995.Garrouch, A.A., Lababidi, H.M.S., and Gharbi, R.B., Dielectric dispersion of dilute suspensions of

colloid particles: Practical applications, Journal of Physical Chemistry, 100, 16996, 1996.Garside, J., General principles of crystallization, in Food Structure and Behaviour, Blanshard, J.M.V.

and Lillford, P., Eds., Academic Press, London, 1987, chap. 3.Garti, N. and Sato, K., Crystallization and Polymorphism of Fats and Fatty Acids, Marcel Dekker, New

York, 1988.Gelin, B.R., Molecular Modeling of Polymer Structures and Properties, Hanser Publishers, Munich,

1994.Glicksman, M., Background and classification, in Food Hydrocolloids, Vol. I, Glicksman, M., Ed.,

CRC Press, Boca Raton, FL, 1982a, chap. 1.Glicksman, M., Functional properties, in Food Hydrocolloids, Vol, I, Glicksman, M., Ed., CRC Press,

Boca Raton, FL, 1982b, chap. 3.Glicksman, M., Food applications of gums, in Food Carbohydrates, Lineback, D.R. and Inglett, G.E.,

Eds., AVI Publishing, Westport, CT, 1982c, chap. 15.Goff, H.D., Interactions and contributions of stabilizers and emulsifiers to development of structure in

ice-cream, in Food Colloids and Polymers: Stability and Mechanical Properties, Dickinson, E.and Walstra, P., Eds., Royal Society of Chemistry, Cambridge, 1993, 71.

Goff, H.D., Liboff, M., Jordon, W.K., and Kinsella, J.E., The effect of polysorbate on the fat emulsionof ice-cream mix: Evidence from transmission electron microscopy studies, Food Microstruc-ture, 6, 193, 1987.

Goodwin, J.W. and Ottewill, R.H., Properties of concentrated colloidal dispersions, Journal of theChemical Society, Faraday Transactions, 87, 357, 1991.

Gopal, E.S.R., Principles of emulsion formation, in Emulsion Science, Sherman, P., Ed., AcademicPress, London, 1968, chap. 1.

Gregory, J., The calculation of Hamaker constants, Advances in Colloid and Interface Science, 2, 396,1969.

Gregory, J., Approximate expressions for retarded van der Waals interaction, Journal of ColloidScience, 83, 138, 1981.

Griffin, W.G. and Griffin, M.C.A., The attenuation of ultrasound in aqueous suspensions of caseinmicelles from bovine milk, Journal of the Acoustical Society of America, 87, 2541, 1990.

Grover, G.S. and Bike, S.G., Monitoring of flocculation in-situ in sterically stabilized silica dispersionsusing rheological techniques, Langmuir, 11, 1807, 1995.

Guichard, E., Interactions between pectins and flavor compounds in strawberry jam, in Flavor–FoodInteractions, McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Washington,D.C., 1996, chap. 11.

Gunstone, F.D. and Norris, F.A., Lipids in Foods: Chemistry, Biochemistry and Technology, PergamonPress, Oxford, 1983.


Gurr, E., Encyclopedia of Microscopic Stains, Williams and Wilkins, Baltimore, 1961.Guyot, C., Bonnafont, C., Lesschaeve, I., Issanchou, S., Voilley, A., and Spinnler, H.E., Effect of fat

content on odor intensity of three aroma compounds in model emulsions: δ-Decalactone,diacetyl and butyric acid, Journal of Agricultural and Food Chemistry, 44, 2341, 1996.

Hallet, F.R., Particle size analysis by dynamic light scattering, Food Research International, 27, 195,1994.

Halliwell, B. and Gutteridge, J.M.C., Free Radicals in Biology and Medicine, 2nd ed., Clarendon Press,London, 1991.

Hamilton, R.J. and Bhati, A., Recent Advances in Chemistry and Technology of Fats and Oils, ElsevierApplied Science, London, 1987.

Hansen, A.P. and Booker, D.C., Flavor interactions with casein and whey protein, in Flavor–FoodInteractions, McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Washington,D.C., 1996, chap. 7.

Harper, W.J. and Hall, C.W., Dairy Technology and Engineering, AVI Publishing, Westport, CT, 1976.Harrison, M., Hills, B.P., Bakker, J., and Clothier, T., Mathematical models of flavor release from liquid

emulsions, Journal of Food Science, 62, 653, 1997.Harvey, B.A., Druaux, C., and Voilley, A., Effect of protein on the retention and transfer of aroma

compounds at the lipid water interface, in Food Macromolecules and Colloids, Dickinson, E.and Lorient, D., Eds., Royal Society of Chemistry, Cambridge, 1995, 154.

Harwalker, V.R. and Ma, C.Y., Thermal Analysis of Foods, Elsevier Science, New York, 1990.Hasenhuettl, G.L., Overview of food emulsifiers, in Food Emulsifiers and Their Applications, Hasenhuettl,

G.L. and Hartel, R.W., Eds., Chapman and Hall, New York, 1997.Hasenhuettl, G.L. and Hartel, R.W., Eds., Food Emulsifiers and Their Applications, Chapman and Hall,

New York, 1997.Hasted, J.B., Liquid water: Dielectric properties, in Water: A Comprehensive Treatise, Vol. 1, Franks,

F., Ed., Plenum Press, New York, 1972, chap. 7.Hato, M., Maruta, M., and Yoshida, T., Surface forces between protein A adsorbed mica surfaces,

Colloids and Surface A, 109, 345, 1996.Hau, M.Y.M., Gray, D.A., and Taylor, A.J., Binding of volatiles to starch, in Flavor–Food Interactions,

McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Washington, D.C., 1996,chap. 10.

Hauser, H., Lipids, in Water: A Comprehensive Treatise, Vol. 4, Franks, F., Ed., Plenum Press, NewYork, 1975, chap. 4.

Heertje, I. and Paques, M., Advances in electron microscopy, in New Physicochemical Techniques forthe Characterization of Complex Food Systems, Dickinson, E., Ed., Blackie Academic andProfessional, London, 1995, chap. 1.

Hemar, Y., Herrmann, N., Lemarechal, P., Hocquart, R., and Lequeux, F., Effective medium model forultrasonic attenuation due to the thermo-elastic effect in concentrated emulsions, Journal ofPhysique II, 7, 637, 1997.

Hermansson, A.M., Gel structure of food biopolymers, in Food Structure — Its Creation and Evalu-ation, Blanshard, J.M.V. and Mitchell, J.R., Eds., Butterworths, London, 1988, chap. 3.

Hernandez, E. and Baker, R.A., Turbidity of beverages with citrus oil clouding agent, Journal of FoodScience, 56, 1024, 1991.

Hernandez, E., Baker, R.A., and Crandall, P.G., Model for evaluating the turbidity in cloudy beverages,Journal of Food Science, 56, 747, 1991.

Hernqvist, L., On the structure of triglycerides in the liquid state and fat crystallinization, Fette SeifenAnstrichmittel, 86, 297, 1984.

Hernqvist, L., Polymorphism of triglycerides: A crystallographic review, Food Structure, 9, 39, 1990.Hiemenz, P.C., Principles of Colloid and Surface Chemistry, Dekker, New York, 1986.Hollingsworth, P., Food research: Cooperation is the key, Food Technology, 49(2), 67, 1995.Holt, W.J.C. and Chan, D.Y.C., Pair interactions between heterogeneous spheres, Langmuir, 13, 1577,

1997.Horne, D.S., Light scattering studies of colloid stability and gelation, in New Physicochemical Tech-

niques for the Characterization of Complex Food Systems, Dickinson, E., Ed., Blackie Aca-demic and Professional, London, 1995, chap. 11.



Huang, S.W., Frankel, E.N., and German, J.B., Antioxidant activity of α- and γ-tocopherols in bulk oilsand oil-in-water emulsions, Journal of Agricultural and Food Chemistry, 42, 2108, 1994.

Huang, S.W., Frankel, E.N., Schwarz, K., Aeschbach, R., and German, J.B., Antioxidant activity ofcarnosic acid and methyl carnosate in bulk oils and oil-in-water emulsions, Journal of Agricul-tural and Food Chemistry, 44, 2951, 1996a.

Huang, S.W., Frankel, E.N., Schwarz, K., and German, J.B., Effect of pH on antioxidant activity ofalpha tocopherol and trolox in oil-in-water emulsions, Journal of Agricultural and FoodChemistry, 44, 2496, 1996b.

Huang, S.W., Hopia, A., Schwarz, K., Frankel, E.N., and German, J.B., Antioxidant activity of alphatocopherol and trolox in different lipid substrates — Bulk oils versus oil-in-water emulsions,Journal of Agricultural and Food Chemistry, 44, 444, 1996c.

Huang, S.W., Frankel, E.N., Aeschabach, R., and German, J.B., Partition of selected antioxidants incorn oil-in-water model systems, Journal of Agricultural and Food Chemistry. 45, 1991, 1997.

Hunt, J.A. and Dalgleish, D.G., Effect of pH on the stability and surface composition of emulsions madewith whey protein isolate, Journal of Food Science, 59, 2131, 1994.

Hunt, J.A. and Dalgleish, D.G., Heat stability of oil-in-water emulsions containing milk proteins: Effectof ionic strength and pH, Journal of Food Science, 60, 1120, 1995.

Hunt, J.A. and Dalgleish, D.G., The effect of the presence of KCl on the adsorption behavior of whey-protein and caseinate in oil-in-water emulsions, Food Hydrocolloids, 10, 159, 1996.

Hunter, R.J., Foundations of Colloid Science, Vol. 1, Oxford University Press, Oxford, 1986.Hunter, R.J., Foundations of Colloid Science, Vol. 2, Oxford University Press, Oxford, 1989.Hunter, R.J., Introduction to Modern Colloid Science, Oxford University Press, Oxford, 1993.Hunter, R.J., The electroacoustic characterization of colloidal suspensions, in Handbook on Ultrasonic

and Dielectric Characterization Techniques for Suspended Particles, Hackley, V.A. and Texter,J., Eds., American Chemical Society, Westerville, OH, 1998.

Hussin, A.B.B.H. and Povey, M.J.W., A study of dilation and acoustic propagation in solidifying fatsand oils: Experimental, Journal of the American Oil Chemists Society, 61, 560, 1984.

Hutchings, J.B., Food Colour and Appearance, Blackie Academic and Professional, London, 1994.Isaacs, E.E., Huang, M., and Babchin, A.J., Electroacoustic method for monitoring the coalescence of

water-in-oil emulsions, Colloids and Surfaces, 46, 177, 1990.Isengard, H.D., Rapid water determination in food stuffs, Trends in Food Science and Technology, 6,

155, 1995.Israelachvili, J.N., Intermolecular and Surface Forces, Academic Press, London, 1992.Israelachvili, J.N., The science and application of emulsions — An overview, Colloids and Surfaces,

91, 1, 1994.Israelachvili, J.N. and Berman A., Irreversibility, energy dissipation and time effects in intermolecular

and surface interactions, Israel Journal of Chemistry, 35, 85, 1995.Israelachvili, J.N. and Pashly, R.M., Measurement of the hydrophobic interaction between two hydro-

phobic surfaces in aqueous electrolyte solutions, Journal of Colloid Science, 98, 500, 1984.Israelachvili, J.N. and Wennerstrom, H., Role of hydration and water-structure in biological and

colloidal interactions, Nature, 379, 219, 1996.Jackel, V.K., Uber die Funktionen des Schuzkolloids, Kolloid-Zeitschrift und Zeitshrift fur Polymere,

197, 143, 1964.Janssen, J.J.M., Boon, A., and Agterof, W.G.M., Influence of dynamic interfacial properties on droplet

breakup in simple shear flocs, Fluid Mechanics and Transport Phenomena, 40, 1929, 1994.Jaynes, E.N., Applications in the food industry. II, in Encyclopedia of Emulsion Technology, Vol. 2,

Becher, P., Ed., Marcel Dekker, New York, 1983, chap. 6.Jenkins, P. and Snowden, M., Depletion flocculation in colloidal dispersions, Advances in Colloid and

Interface Science, 68, 57, 1996.Jokela, P., Fletcher, P.D.I., Aveyard, R., and Lu, J.R., The use of computerized microscopic image

analysis to determine droplet size distributions, Journal of Colloid and Interface Science, 134,417, 1990.

Jost, R., Baechler, R., and Masson, G., Heat gelation of oil-in-water emulsions stabilized by wheyprotein, Journal of Food Science, 51, 440, 1986.

Jungermann, E., Fat based surface-active agents, in Bailey’s Industrial Oil and Fat Products, Vol. 1,Swern, D., Ed., John Wiley & Sons, New York, 1979, chap. 9.


Kabalnov, A.S. and Shchukin, E.D., Ostwald ripening theory: Applications to fluorocarbon emulsionstability, Advances in Colloid and Interface Science, 38, 69, 1992.

Kabalnov, A. and Weers, J., Macroemulsion stability within the Winsor III region: Theory versusexperiment, Langmuir, 12, 1931, 1996.

Kabalnov, A. and Wennerstrom, H., Macroemulsion stability: The orientated wedge theory revisited,Langmuir, 12, 276, 1996.

Kabalnov, A., Weers, J., Arlauskas, R., and Tarara, T., Phospholipids as emulsion stabilizers. 1.Interfacial tension, Langmuir, 11, 2966, 1995.

Kalab, M., Electron microscopy of milk products: A review of techniques, Scanning Electron Micros-copy, 3, 453, 1981.

Kalab, M., Allan-Wojtas, P., and Miller, S.S., Microscopy and other imaging techniques in foodstructure analysis, Trends in Food Science and Technology, 6, 180, 1995.

Kallay, N., Hlady, V., Jednacak-Biscan, J., and Milonjic, S., Techniques for the study of adsorptionfrom solution, in Physical Methods of Chemistry, Vol. IXA, Rossiter, B.W. and Baetzold, R.C.,Eds., John Wiley & Sons, New York, 1993, chap. 2.

Kandori, K., Applications of microporous glass membranes: Membrane emulsification, in Food Pro-cessing: Recent Developments, Gaonkar, A.G., Ed., Elsevier Science, Amsterdam, 1995.

Karbstein, H. and Schubert, H., Developments in the continuous mechanical production of oil-in-watermacro-emulsions, Chemical Engineering and Processing, 34, 205, 1995.

Karplus, M. and Porter, R.N., Atoms and Molecules: An Introduction for Students of Physical Chem-istry, Benjamin-Cummins, Menlo Park, CA, 1970.

Kato, A., Fujishige, T., Matsudomi, N., and Kogayashi, K., Determination of emulsifying properties ofsome proteins by conductivity measurements, Journal of Food Science, 50, 56, 1985.

Katz, F., Technology trends, Food Technology, 51(6), 46, 1997.Kawano, S., Progress in application of NIR and FT-IR in food characterization, in Characterization of

Foods: Emerging Techniques, Gaonkar, A.G., Ed., Elsevier, Amsterdam, 1995, chap. 8.Keikens, F., Optimization of electrical conductance measurements for the quantification and prediction

of phase separation in O/W emulsions containing hydroxpropylmethylcelluloses as emulsifyingagents, International Journal of Pharmaceutics, 146, 239, 1997.

Kellens, M., Meeussen, W., and Reynaers, H., Study of the polymorphism and the crystallizationkinetics of tripalmitin, Journal of the American Oil Chemists Society, 69, 906, 1992.

Kerker, M., The Scattering of Light and Other Electromagnetic Radiation, Academic Press, New York,1969.

Kern, C.W. and Karplus, M., The water molecule, in Water: A Comprehensive Treatise, Vol. 1, Franks,F., Ed., Plenum Press, New York, 1972, chap. 2.

Kessler, H.G., Food Engineering and Dairy Technology, Verlag A., Kessler Publishing, Freising,Germany, 1981.

Kinsella, J.E., Relationships between structure and functional properties of food proteins, in FoodProteins, Fox, P.F. and Condon, J.J., Eds., Applied Science Publishers, London, 1982, 51.

Kinsella, J.E., in Flavor Chemistry of Lipid Foods, Min, D.B. and Smouse, T.H., American OilChemists Society, Champaign, IL, 1989, 376.

Kinsella, J.E. and Whitehead, D.M., Proteins in whey: Chemical, physical and functional properties,Advances in Food and Nutrition Research, 33, 343, 1989.

Kirby, A.R., Gunning, A.P., and Morris, V.J., Atomic force microscopy in food research: A newtechnique comes of age, Trends in Food Science and Technology, 6, 359, 1995.

Kitakara, A. and Watanabe, A., Electrical Phenomenon at Interfaces: Fundamentals, Measurementsand Applications, Marcel Dekker, New York, 1984.

Klemaszeski, J., Haque, Z., and Kinsella, J.E., An electronic imaging system for determining the dropletsize and dynamic breakdown of protein stabilized emulsions, Journal of Food Science, 54, 440,1989.

Kokini, J.L., The physical basis of liquid food texture and texture–taste interactions, Journal of FoodEngineering, 6, 51, 1987.

Kokini, J.L., Eads, T., and Ludescher, R.D., Research needs on the molecular basis for food function-ality, Food Technology, 47(3), 36S, 1993.

Kresheck, G.C., Surfactants, in Water: A Comprehensive Treatise, Vol. 4, Franks, F., Ed., PlenumPress, New York, 1975, chap. 2.



Krog, M.J., Riisom, T.H., and Larsson, K., Applications in the food industry. I, in Encyclopedia ofEmulsion Technology, Vol. 2, Becker, P., Ed., Marcel Dekker, New York, 1983, chap. 5.

Kumosinski, T.F., Brown, E.M., and Farrell, H.M., Molecular modeling in food research: Technologyand techniques, Trends in Food Science and Technology, 2, 110, 1991a.

Kumosinski, T.F., Brown, E.M., and Farrell, H.M., Molecular modeling in food research: Applications,Trends in Food Science and Technology, 2, 190, 1991b.

Lachaise, J., Mendiboure, B., Dicharry, C., Marion, G., and Salager, J.L., Simulation of theoveremulsification phenomenon in turbulent stirring, Colloids and Surfaces A, 110, 1, 1996.

Land, D.G., Perspectives on the effects of interactions on flavour perception: An overview, in Flavor–Food Interactions, McGorrin, R.J. and Leland, J.V., Eds., American Chemical Society, Wash-ington, D.C., 1996.

Land, J. and Zana, R., Chemical relaxation methods, in Surfactant Solutions: New Methods of Inves-tigation, Zana, R., Ed., Marcel Dekker, New York, 1987.

Landy, P., Courthaudon, J.L., Dubois, C., and Voilley, A., Effect of interface in model food emulsionson the volatility of aroma compounds, Journal of Agricultural and Food Chemistry, 44, 526,1996.

Langourieux, S. and Crouzet, J., Protein–aroma interactions, in Food Macromolecules and Colloids,Dickinson, E. and Lorient, D., Eds., Royal Society of Chemistry, Cambridge, 1995, 123.

Lapasin, R. and Pricl, S., Rheology of Industrial Polysaccharides: Theory and Applications, BlackieAcademic and Professional, London, 1995.

Larison, K.D., Handbook of Fluorescent Probes and Research Chemicals, Molecular Probes, Eugene,OR, 1992.

Launay, B., Doublier, J.L., and Cuvelier, G., Flow properties of aqueous solutions of polysaccharides,in Functional Properties of Food Macromolecules, Mitchell, J.R. and Ledward, D.A., Eds.,Elsevier, London, 1986, chap. 1.

Lawson, H.W., Food Oils and Fats: Technology, Utilization and Nutrition, Chapman and Hall, NewYork, 1995.

Ledward, D.A., Gelation of gelatin, in Functional Properties of Food Macromolecules, Mitchell, J.R.and Ledward, D.A., Eds., Elsevier Applied Science, London, 1986, chap. 4.

Lee, S.Y. and Morr, C.V., Fixation and staining milkfat globules in cream for transmission and scanningelectron microscopy, Journal of Food Science, 57, 887, 1992.

Lees, L.H. and Pandolfe, W.D., Homogenizers, in Encyclopedia of Food Engineering, Hall, C.W.,Farrall, A.W., and Rippen, A.L., Eds., AVI Publishing, Westport, CT, 1986, 467.

Lehnert, S., Tarabishi, H., and Leuenberger, H., Investigation of thermal phase inversion in emulsions,Colloids and Surfaces A, 91, 227, 1994.

Lehninger, A.L., Nelson, D.L., and Cox, M.M., Principles of Biochemistry, 2nd ed., Worth Publishers,New York, 1993.

Levine, H. and Slade, L., Water Relationships in Foods: Advances in the 1980’s and Trends for the1990’s, Plenum Press, New York, 1991.

Leward, D.A., Gelation of gelatin, in Functional Properties of Food Macromolecules, Mitchell, J.R.and Ledward, D.A., Eds., Elsevier Applied Science, London, 1986, chap. 4.

Li, X., Cox, J.C., and Flumerfelt, R.W., Determination of emulsion size distribution by NMR restricteddiffusion measurement, AIChE Journal, 38, 1671, 1992.

Lickiss, P.D. and McGrath, V.E., Breaking the sound barrier, Chemistry in Britain, 32, 47, 1996.Lindman, B., Carlsson, A., Gerdes, S., Karlstrom, G., Picullel, L., Thalberg, K., and Zhang, K.,

Polysaccharide–surfactant systems: Interactions, phase diagrams and novel gels, in Food Col-loids and Polymers: Stability and Mechanical Properties, Dickinson, E. and Walstra, P., Eds.,Royal Society of Chemistry, Cambridge, 1993, 113.

Linfield, W.M., Anionic Surfactants: Parts I and II, Marcel Dekker, New York, 1976.Lips, A., Campbell, I.J., and Pelan, E.G., Aggregation mechanisms in food colloids and the role of

biopolymers, in Food Polymers, Gels and Colloids, Dickinson, E., Ed., Royal Society ofChemistry, Cambridge, 1991, 1.

Lips, A., Westbury, T., Hart, P.M., Evans, I.D., and Campbell, I.J., Aggregation mechanisms in foodcolloids and the role of biopolymers, in Food Colloids and Polymers: Stability and MechanicalProperties, Dickinson, E. and Walstra, P., Eds., Royal Society of Chemistry, Cambridge, 1993,31.


Lissant, K.J., Demulsification: Industrial Applications, Marcel Dekker, New York, 1983.Liu, S. and Masliyah, J.H., Rheology of suspensions, in Suspensions: Fundamentals and Applications

in the Petroleum Industry, Schramm, L.L., Ed., American Chemical Society, Washington, D.C.,1996, chap. 3.

Lloyd, N.E., Determination of surface-average particle diameter of colored emulsions by reflectance,and application to emulsion stability studies, Journal of Colloid and Interface Science, 14, 441,1959.

Loncin, M. and Merson, R.L., Food Engineering: Principles and Selected Applications, AcademicPress, New York, 1979.

Lovsey, S.W., Theory of Neutron Scattering from Condensed Matter, Clarendon Press, Oxford, 1984.Lucassen-Reyanders, E.H. and Kuijpers, K.A., The role of interfacial properties in emulsification,

Colloids and Surfaces, 65, 175, 1992.Luckham, P.F. and Costello, B.A. de L., Recent developments in the measurement of interparticle

forces, Advances in Colloid and Interface Science, 44, 193, 1993.Ludescher, R.D., Molecular dynamics of food proteins: Experimental techniques and observations,

Trends in Food Science and Technology, 1, 145, 1990.Luyten, H., Jonkman, M., Kloek, W., and van Vliet, T., Creaming behaviour of dispersed particles in

dilute xanthan solutions, in Food Colloids and Polymers: Stability and Mechanical Properties,Dickinson, E. and Walstra, P., Eds., Royal Society of Chemistry, Cambridge, 1993, 224.

Ma, Y., Varadan, V.K., and Varadan, V.V., Comments on ultrasonic propagation in suspensions,Journal of the Acoustical Society of America, 87, 2779, 1990.

MacGregor, E.A. and Greenwood, C.T., Polymers in Nature, John Wiley, New York, 1980.MacKay, R.A., Solubilization, in Nonionic Surfactants: Physical Chemistry, Schick, M.J., Ed., Marcel

Dekker, New York, 1987, 297.Macosko, C.W., Rheology: Principles, Measurements and Applications, VCH Publishers, New York,

1994.Magdassi, S., Surface Activity of Proteins: Chemical and Physicochemical Modifications, Marcel

Dekker, New York, 1996.Magdassi, S. and Kamyshny, A., Surface activity and functional properties of proteins, in Surface

Activity of Proteins: Chemical and Physicochemical Modifications, Magdassi, S., Ed., MarcelDekker, New York, 1996, chap. 1.

Mahanty, J. and Ninham, B.W., Dispersion Forces, Academic Press, New York, 1976.Malmsten, M., Lindstrom, A.-L., and Warnheim, T., Ellipsometry studies of interfacial film formation

in emulsion systems, Journal of Colloid and Interface Science, 173, 297, 1995.Marra, J., Direct measurements of attractive van der Waals and adhesion forces between uncharged lipid

bilayers in aqueous solutions, Journal of Colloid and Interface Science, 109, 11, 1986.Matsubara, T. and Texter, J., In situ voltammetric determinations of solute distribution coefficients in

emulsions, Journal of Colloid and Interface Science, 112, 421, 1986.Matthew, J.B., Electrostatic effects in proteins, Annual Review in Biophysics and Biophysical Chem-

istry, 14, 387, 1985.Mayhill, P.G. and Newstead, D.F., The effect of milkfat fractions and emulsifier type on creaming in

normal-solids UHT recombined milk, Milchwissenschaft, 47, 75, 1992.McCarthy, W.W., Ultrasonic emulsification, Drug and Cosmetic Industry, 94(6), 821, 1964.McClements, D.J., The Use of Ultrasonics for Characterizing Fats and Emulsions, Ph.D. thesis, Uni-

versity of Leeds, United Kingdom, 1989.McClements, D.J., Ultrasonic characterization of emulsions and suspensions, Advances in Colloid and

Interface Science, 37, 33, 1991.McClements, D.J., Ultrasonic determination of depletion flocculation in oil-in-water emulsions contain-

ing a non-ionic surfactant, Colloids and Surfaces A, 90, 25, 1994.McClements, D.J., Principles of ultrasonic droplet size determination, Langmuir, 12, 3454, 1996.McClements, D.J. and Dungan, S.R., Factors that affect the rate of oil exchange between oil-in-water

emulsion droplets stabilized by a non-ionic surfactant: Droplet size, surfactant concentrationand ionic strength, Journal of Physical Chemistry, 97, 7304, 1993.

McClements, D.J. and Dungan, S.R., Effect of colloidal interactions on the rate of interdroplet hetero-geneous nucleation in oil-in-water emulsions, Journal of Colloid and Interface Science, 186,17, 1997.



McClements, D.J. and Fairley, P., Ultrasonic pulse echo reflectometer, Ultrasonics, 29, 58, 1991.McClements, D.J. and Fairley, P., Frequency scanning ultrasonic pulse echo reflectometer, Ultrasonics,

30, 403, 1992.McClements, D.J. and Keogh, M.K., Physical properties of cold-setting gels prepared from heat-

denatured whey protein isolate, Journal of Food Chemistry and Agriculture, 69, 7, 1995.McClements, D.J. and Povey, M.J.W., Solid fat content determination using ultrasonic velocity mea-

surements, International Journal of Food Science and Technology, 22, 491, 1987.McClements, D.J. and Povey, M.J.W., Comparison of pulsed NMR and ultrasonic velocity techniques

for determining solid fat contents, International Journal of Food Science and Technology, 23,159, 1988.

McClements, D.J. and Povey, M.J.W., Scattering of ultrasound by emulsions, Journal of Physics D:Applied Physics, 22, 38, 1989.

McClements, D.J., Povey, M.J.W., Jury, M., and Betsansis, E., Ultrasonic characterization of a foodemulsion, Ultrasonics, 28, 266, 1990.

McClements, D.J., Dungan, S.R., German, J.B., Simoneau, C., and Kinsella, J.E., Droplet size andemulsifier type affect crystallization and melting of hydrocarbon-in-water emulsions, Journalof Food Science, 58, 1148, 1993a.

McClements, D.J., Dungan, S.R., German, J.B., and Kinsella, J.E., Factors which affect oil exchangebetween oil-in-water emulsion droplets stabilized by whey protein isolate: Protein concentra-tion, droplet size and ethanol, Colloids and Surfaces A, 81, 203, 1993b.

McClements, D.J., Monahan, F.J., and Kinsella, J.E., Effect of emulsion droplets on the rheology ofwhey protein isolate gels, Journal of Texture Studies, 24, 411, 1993c.

McClements, D.J., Monahan, F.J., and Kinsella, J.E., Disulfide bond formation affects the stability ofwhey protein stabilized emulsion, Journal of Food Science, 58, 1036, 1993d.

McClements, D.J., Dickinson, E., Dungan, S.R., Kinsella, J.E., Ma, J.E., and Povey, M.J.W., Effect ofemulsifiers type on the crystallization kinetics of oil-in-water emulsions containing a mixtureof solid and liquid droplets, Journal of Colloid and Interface Science, 160, 293, 1993e.

McClements, D.J., Povey, M.J.W., and Dickinson, E., Absorption and velocity dispersion due tocrystallization and melting of emulsion droplets, Ultrasonics, 31, 443, 1993f.

McClements, D.J., Han, S.W., and Dungan, S.R., Interdroplet heterogeneous nucleation of supercooledliquid droplets by solid droplets in oil-in-water emulsions, Journal of the American Oil Chem-ists Society, 71, 1385, 1994.

McClements, D.J., Chantrapornchai, W., and Clydesdale, F., Prediction of food emulsion color usinglight scattering theory, Journal of Food Science, submitted, 1998a.

McClements, D.J., Hemar, Y., and Herrmann, N., Incorporation of thermal overlap effects into multiplescattering theory, Journal of the Acoustical Society of America, submitted, 1998b.

McNulty, P.B., Flavor release — Elusive and dynamic, in Food Structure and Behaviour, Blanshard,J.M.V. and Lillford, P., Eds., Academic Press, London, 1987.

McNulty, P.B. and Karel, M., Factors affecting flavour release and uptake in O/W emulsions. II. Stirredcell studies, Journal of Food Technology, 8, 319, 1973.

Mei, L., McClements, D.J., Wu, J., and Decker, E.A., Iron-catalyzed lipid oxidation in emulsions asaffected by surfactant, pH and NaCl, Food Chemistry, 61, 307, 1998.

Melik, D.H. and Fogler, H.S., Fundamentals of colloidal stability in quiescent media, in Encyclopediaof Emulsion Technology, Vol. 3, Becher, P., Marcel Dekker, New York, 1988, chap. 1.

Menon, W.B. and Wasan, D.T., Demulsification, in Encyclopedia of Emulsion Technology, Vol. 2,Becher, P., Ed., Marcel Dekker, New York, 1985, chap. 1.

Mewis, J. and Macosko, C.W., Suspension rheology, in Rheology: Principles, Measurements andApplications, Macosko, C.W., Ed., VCH Publishers, New York, 1994, chap. 10.

Miklavic, S.J. and Ninham, B.W., Competition for adsorption sites by hydrated ions, Journal of Colloidand Interface Science, 134, 305, 1990.

Mikula, R.J., Emulsion characterization, in Emulsions: Fundamentals and Applications in the Petro-leum Industry, Schramm, L.L., Ed., American Chemical Society, Washington, D.C., 1992,chap. 3.

Miles, M.J. and McMaster, T.J., Scanning probe microscopy of food-related systems, in New Physico-chemical Techniques for the Characterization of Complex Food Systems, Dickinson, E., Ed.,Blackie Academic and Professional, London, 1995, chap. 3.


Miles, C.A., Fursey, G.A.J., and Jones, C.D.J., Ultrasonic estimation of solid/liquid ratios in fats, oilsand adipose tissue, Journal of the Science of Food and Agriculture, 36, 215, 1985.

Miles, C.A., Shore, A., and Langley, K.R., Attenuation of ultrasound in milks and creams, Ultrasonics,28, 394, 1990.

Miller, C.A. and Neogi, P., Interfacial Phenomena, Marcel Dekker, New York, 1985.Milling, A., Mulvaney, P., and Larson, E., Direct measurement of repulsive van der Waals interactions

using an atomic force microscope, Journal of Colloid and Interface Science, 180, 460, 1996.Min, D.B., Crude fat analysis, in Introduction to the Chemical Analysis of Foods, Nielsen, S.S., Ed.,

Bartlett and Jones, Boston, 1994, chap. 12.Mishchuk, N.A., Sjoblom, J., and Dukhin, S.S., Influence of retardation and screening on van der Waals

attractive forces on reverse coagulation of emulsions in the secondary minimum, ColloidJournal, 57, 785, 1995.

Mishchuk, N.A., Sjoblom, J., and Dukhin, S.S., The effect of retardation and screening of van der Waalsattractive forces on the breaking of doublet of drops during sedimentation, Colloid Journal, 58,210, 1996.

Mitchell, J.R. and Ledward, D.A., Functional Properties of Food Macromolecules, Elsevier AppliedScience, London, 1986.

Mittal, K.L., Solution Chemistry of Surfactants, Vol. 1, Plenum Press, New York, 1979.Monahan, F.J., McClements, D.J., and German, J.B., Disulfide-mediated polymerization reactions and

physical properties of heated WPI-stabilized emulsions. Journal of Food Science, 61, 504, 1996.Moran, D.P.J., Fats in spreadable products, in Fats in Food Products, Blackie Academic and Profes-

sional, London, 1994.Moran, D.P.J. and Rajah, K.K., Fats in Food Products, Blackie Academic and Professional, London,

1994.Morris, V.J., Gelation of polysaccharides, in Functional Properties of Food Macromolecules, Mitchell,

J.R. and Ledward, D.A., Eds., Elsevier Applied Science, London, 1986, chap. 3.Mukerjee, P., Solubilization in aqueous micellar systems, in Solution Chemistry of Surfactants, Vol. 1,

Mittal, K.L., Ed., Plenum Press, London, 1979.Mulder, H. and Walstra, P., The Milk Fat Globule: Emulsion Science as Applied to Milk Products and

Comparable Foods, Pudoc, Wageningen, Netherlands, 1974.Mullin, J.W., Crystallization, 3rd ed., Butterworth-Heinmann, Oxford, 1993.Mulvihill, D.M. and Kinsella, J.E., Gelation characteristics of whey proteins and β-lactoglobulin, Food

Technology, 41, 102, 1987.Mulvihill, D.M. and Kinsella, J.E., Gelation of β-lactoglobulin: Effects of sodium chloride and calcium

chloride on the rheological and structural properties of gels, Journal of Food Science, 53, 231,1988.

Murray, B.S. and Dickinson, E., Interfacial rheology and the dynamic properties of adsorbed films offood proteins and surfactants, Food Science and Technology International, 2, 131, 1996.

Murrell, J.N. and Boucher, E.A., Properties of Liquids and Solutions, John Wiley & Sons, Chichester,U.K., 1982.

Murrell, J.N. and Jenkins, A.D., Properties of Liquids and Solutions, 2nd ed., John Wiley & Sons,Chichester, U.K., 1994.

Myers, D., Surfactant Science Technology, VCH Publishers, Weinheim, Germany, 1988.Nakai, S. and Li-Chan, E., Hydrophobic Interactions in Food Systems, CRC Press, Boca Raton, FL,

1988.Nakamura, H., Roles of electrostatic interactions in proteins, Quarterly Review of Biophysics, 29, 1,

1996.Nawar, W.W., Lipids, in Food Chemistry, 3rd ed., Fennema, O.R., Ed., Marcel Dekker, New York,

1996, 225.Nielsen, S.S., Introduction to food analysis, in Introduction to the Chemical Analysis of Foods, Nielsen,

S.S., Ed., Jones and Bartlett, Boston, 1994, chap. 1.Ninham, B.W. and Yaminsky, V., Ion binding and ion specificity: The Hofmeister effect and Onsager

and Lifshitz theories, Langmuir, 13, 2097, 1997.O’Brien, R.W., Cannon, D.W., and Rowlands, W.N., Electroacoustic determination of particle size and

zeta potential, Journal of Colloid and Interface Science, 173, 406, 1995.O’Donnell, C.D., Controlling the fat: What’s new, what’s to come, Prepared Foods, p. 65, July 1995.



Oh, S.G., Jobalia, M., and Shah, D.O., The effect of micellar lifetime on the droplet size in emulsions,Journal of Colloid and Interface Science, 155, 511, 1993.

Okshima, H., Electrostatic interaction between two spherical colloidal particles, Advances in Colloidand Interface Science, 53, 77, 1994.

O’Neill, T.E., Flavor binding to food proteins: An overview, in Flavor–Food Interactions, McGorrin,R.J. and Leland, J.V., Eds., American Chemical Society, Washington, D.C., 1996, chap. 6.

Oretega-Vinuesa, J.L., Marin-Rodriguez, A., and Hidalgo-Alvarez, R., Colloidal stability of polymercolloids with different interfacial properties: Mechanisms, Journal of Colloid and InterfaceScience, 184, 259, 1996.

Orr, C., Determination of particle size, in Encyclopedia of Emulsion Technology, Vol. 3, Becher, P.,Ed., Marcel Dekker, New York, 1988, chap. 3.

Overbosch, P., Afterof, W.G.M., and Haring, P.G.M., Flavor release in the mouth, Food ReviewsInternational, 7, 137, 1991.

Ozilgen, S., Simoneau, C., German, J.B., McCarthy, M.J., and Reid, D.S., Crystallization kinetics ofemulsified triglycerides, Journal of the Science of Food and Agriculture, 61, 101, 1993.

Pacek, A.W., Moore, I.P.T., Nienow, A.W., and Calabrese, R.V., Video technique for measuringdynamics of liquid–liquid dispersion during phase inversion, AIChE Journal, 40, 1940, 1994.

Pailthorpe, B.A. and Russel, W.B., Retarded van der Waals interaction between spheres, Journal ofColloid and Interface Science, 89, 563, 1982.

Pal, R., Rheology of polymer thickened emulsions, Journal of Rheology, 36, 1245, 1992.Pal, R., Techniques for measuring the composition (oil and water content) of emulsions — A state of

the art review, Colloids and Surfaces A, 84, 141, 1994.Pal, R., Rheology of emulsions containing polymeric liquids, in Encyclopedia of Emulsion Technology,

Vol. 4, Becher, P., Ed., Marcel Dekker, New York, 1996, chap. 3.Pal, R., Yan, Y., and Masliyah, J.H., Rheology of emulsions, in Emulsions: Fundamentals and Appli-

cations in the Petroleum Industry, Schramm, L.L., Ed., American Chemical Society, Washing-ton, D.C., 1992, chap. 4.

Pandolfe, W.D., Homogenizers, in Encyclopedia of Food Science and Technology, John Wiley & Sons,New York, 1991, 1413.

Pandolfe, W.D., Effect of premix condition, surfactant concentration and oil level on the formation ofoil-in-water emulsions by homogenization, Journal of Dispersion Science and Technology, 16,633, 1995.

Pandolfe, W.D. and Masucci, S.F., An instrument for rapid spectroturbimetric analysis, AmericanLaboratory, p. 40, August 1984.

Parsegian, V.A., Reconciliation of van der Waals force measurements between phosphatidylcholinebilayers in water and between bilayer coated mica surfaces, Langmuir, 9, 3625, 1993.

Partmann, W., The effects of freezing and thawing on food quality, in Water Relationships of Foods,Duckworth, R.B., Ed., Academic Press, London, 1975, 505.

Pashley, R.M., McGuiggan, P.M., Ninham, B.W., and Evans, D.F., Attractive forces between un-charged hydrophobic surfaces: Direct measurements in aqueous solutions, Science, p. 1088, 1985.

Patterson, D., Pigments, Elsevier, Amsterdam, 1967.Paulaitis, M.E., Garde, S., and Ashbaugh, H.S., The hydrophobic effect, Current Opinion in Colloid and

Interface Science, 1, 376, 1996.Pearce, K.N. and Kinsella, J.E., Emulsifying properties of proteins: Evaluation of a turbidimetric

technique, Journal of Agricultural and Food Chemistry, 26, 716, 1978.Peleg, M., Fractals and foods, Critical Reviews in Food Science and Nutrition, 33, 149, 1993.Penner, M.H., Basic principles of spectroscopy, in Introduction to the Chemical Analysis of Foods,

Nielsen, S.S., Ed., Jones and Bartlett, London, 1994a, chap. 22.Penner, M.H., Ultraviolet, visible and fluorescence spectroscopy, in Introduction to the Chemical

Analysis of Foods, Nielsen, S.S., Ed., Jones and Bartlett, London, 1994b, chap. 23.Pettitt, D.J., Waybe, J.E.B., Nantz, J.R., and Shoemaker, C.F., Rheological properties of solutions and

emulsions stabilized with xanthan gum and propylene glycol alginate, Journal of Food Science,60, 528, 1995.

Phipps, L.W., Heterogeneous and homogeneous nucleation in supercooled triglycerides and n-paraffins,Transactions of the Faraday Society, 60, 1873, 1964.


Phipps, L.W., The High Pressure Dairy Homogenizer, The National Institute for Research in Dairying,Reading, England, 1985.

Pike, O.A., Fat characterization, in Introduction to the Chemical Analysis of Foods, Nielsen, S.S., Ed.,Jones and Bartlett, London, 1994, chap. 13.

Pinfield, V.J., Dickinson, E., and Povey, M.J.W., Modeling of concentration profiles and ultrasoundvelocity profiles in a creaming emulsion: Importance of scattering effects, Journal of Colloidand Interface Science, 166, 363, 1994.

Pinfield, V.J., Dickinson, E., and Povey, M.J.W., Modeling of combined creaming and flocculation inemulsions, Journal of Colloid and Interface Science, 186, 80, 1997.

Pomeranz, Y. and Meloan, C.E., Food Analysis, 3rd ed., Chapman and Hall, New York, 1994.Pothakamury, U.E. and Barbosa-Canovas, G.V., Fundamental aspects of controlled release in foods,

Trends in Food Science and Technology, 6, 397, 1995.Povey, M.J.W., Ultrasound studies of shelf-life and crystallization, in New Physicochemical Techniques

for the Characterization of Complex Food Systems, Dickinson, E., Ed., Blackie Academic andProfessional, London, 1995, chap. 9.

Povey, M.J.W., Ultrasonic Techniques for Fluids Characterization, Academic Press, San Diego, 1997.Rabinovich, Y.A.I. and Derjaguin, B.V., Interaction of hydrophobized filaments in aqueous electrolyte

solutions, Colloids and Surfaces, 30, 243, 1988.Race, S.W., Improved product quality through viscosity measurement, Food Technology, 45, 86, 1991.Rao, M.A., Rheological properties of fluid foods, in Engineering Properties of Foods, 2nd ed., Rao,

M.A. and Rizvi, S.S.H., Eds., Marcel Dekker, New York, 1995, chap. 1.Rao, V.N.M., Delaney, R.A.M., and Skinner, G.E., Rheological properties of solid foods, in Engineer-

ing Properties of Foods, 2nd ed., Rao, M.A. and Rizvi, S.S.H., Eds., Marcel Dekker, NewYork, 1995, chap. 2.

Reddy, S.R. and Fogler, H.S., Emulsion stability: Determination from turbidity, Journal of Colloid andInterface Science, 79, 101, 1981.

Reichardt, C., Solvents and Solvent Effects in Organic Chemistry, 2nd ed., VCH Publishers, Weinheim,Germany, 1988.

Reiner, E.S. and Radke, C.J., Double layer interactions between charge regulated surfaces, Advancesin Colloid and Interface Science, 47, 59, 1993.

Rha, C.K. and Pradipasena, P., Viscosity of proteins, in Functional Properties of Food Macromol-ecules, Mitchell, J.R. and Ledward, D.A., Eds., Elsevier, London, 1986, chap. 2.

Richmond, J.M., Cationic Surfactants: Organic Chemistry, Marcel Dekker, New York, 1990.Rizvi, S.S.H., Singh, R.K., Hotchkiss, J.H., Heldman, D.R., and Leung, H.K., Research needs in food

engineering, processing and packaging, Food Technology, 47(3), 268, 1993.Roberts, D.D. and Acree, T.E., Retronasal flavor release in oil and water model systems with an

evaluation of volatility predictors, in Flavor–Food Interactions, McGorrin, R.J. and Leland,J.V., Eds., American Chemical Society, Washington, D.C., 1996, chap. 16.

Roberts, D.D., Elmore, J.S., Langley, K.R., and Bakker, J., Effects of sucrose, guar gum and carboxy-methylcellulose on the release of volatile flavor compounds under dynamic conditions, Journalof Agricultural and Food Chemistry, 44, 1321, 1996.

Robin, O., Britten, M., and Paquin, P., Influence of the disperse phase distribution on the electricalconductivity of liquid O/W model and dairy emulsions, Journal of Colloid and InterfaceScience, 167, 401, 1994.

Robinson, G.W., Zhu, S.-B., Singh, S., and Evans, M.W., Water in Biology, Chemistry and Physics:Experimental Overviews and Computational Methodologies, World Scientific, Singapore,1996.

Rogers, N.K., in Prediction of Protein Structure and the Principles of Protein Conformation, Fasman,G.D., Ed., Plenum Press, New York, 1989, 359.

Roozen, J.P., Frankel, E.N., and Kinsellan, J.E., Enzymic and autooxidation of lipids in low fat foods:Model of linoleic acid emulsion in emulsified hexadecane, Food Chemistry, 50, 33, 1994a.

Roozen, J.P., Frankel, E.N., and Kinsellan, J.E., Enzymic and autooxidation of lipids in low fat foods:Model of linoleic acid emulsion in emulsified triolein and vegetable oils, Food Chemistry, 50,39, 1994b.

Rosen, M.J., Surfactants and Interfacial Phenomenon, Wiley-Interscience, New York, 1978.



Roth, C.M. and Lenhoff, A.M., Improved parametric representation of water dielectric data for Lifshitztheory calculations, Journal of Colloid and Interface Science, 179, 637, 1996.

Sader, J.E., Carnie, S.L., and Chan, D.Y.C., Accurate analytic formulas for the double-layer interactionbetween spheres, Journal of Colloid and Interface Science, 171, 4, 1995.

Salager, J.L., Phase transformation and emulsion inversion on the basis of the catastrophe theory, inEncyclopedia of Emulsion Technology, Vol. 3, Becher, P., Marcel Dekker, New York, 1988,chap. 1.

Salvador, D., Bakker, J., Langley, K.R., Potjewijd, R., Martin, A., and Elmore, S., Flavor release ofdiacetyl from water, sunflower oil and emulsions in model systems, Food Quality Preference,5, 103, 1994.

Sand, R., Structure and conformation of hydrocolloids, in Food Hydrocolloids, Vol. I, Glicksman, M.,Ed., CRC Press, Boca Raton, FL, 1982, chap. 2.

Sato, K., Crystallization of fats and fatty acids, in Crystallization and Polymorphism of Fats and FattyAcids, Garti, N. and Sato, K., Eds., Marcel Dekker, New York, 1988, 227.

Schenkel, J.H. and Kitchner, J.A., A test of the Derjaguin–Verwey–Overbeek theory with a colloidalsuspension, Transactions of the Faraday Society, 56, 161, 1960.

Schmidt, R.K., Tasaki, K., and Brady, J.W., Computer modeling studies of the interaction of water withcarbohydrates, Journal of Food Engineering, 22, 43, 1994.

Schubert, H. and Armbruster, H., Principles of formation and stability of emulsions, InternationalChemical Engineering, 32, 14, 1992.

Schultz, K.W., Day, E.A., and Sinnbuber, R.O., Lipids and Their Oxidation, AVI Publishing, Westport,CT, 1962.

Sears, F.W. and Salinger, G.L., Thermodynamics, Kinetic Theory and Statistical Thermodynamics,Addison-Wesley, Reading, MA, 1975.

Seebergh, J.E. and Berg, J.C., Depletion flocculation of aqueous, electrostatically stabilized latexdispersions, Langmuir, 10, 454, 1994.

Semenova, M.G., Factors determining the character of biopolymer–biopolymer interactions in multi-component aqueous solutions modeling food systems, in Macromolecular Interactions in FoodTechnology, Parris, N., Kato, A., Creamer, L.K., and Pearce, J., Eds., American ChemicalSociety, Washington, D.C., 1996, chap. 3.

Shaw, D.J., Introduction to Colloid and Surface Chemistry, 3rd ed., Butterworths, London, 1980.Sherman, P., Ed., Emulsion Science, Academic Press, London, 1968a.Sherman, P., General properties of emulsions and their constituents, in Emulsion Science, Sherman, P.,

Ed., Academic Press, London, 1968b, chap. 3.Sherman, P., Rheology of emulsions, in Emulsion Science, Sherman, P., Ed., Academic Press, London,

1968c, chap. 4.Sherman, P., Industrial Rheology with Particular Reference to Foods, Pharmaceuticals and Cosmetics,

Academic Press, London, 1970.Sherman, P., Rheology of emulsions, in Encyclopedia of Emulsion Technology, Basic Theory, Vol. 1,

Becher, P., Ed., Marcel Dekker, New York, 1982, chap. 7.Sherman, P., A critique of some methods proposed for evaluating the emulsifying capacity and emul-

sion stabilizing performance of vegetable proteins, Italian Journal of Food Science, 1, 3, 1995.Shinoda, K. and Friberg, S., Emulsions and Solubilization, John Wiley & Sons, New York, 1986.Shinoda, K. and Kunieda, H., Phase properties of emulsions: PIT and HLB, in Encyclopedia of

Emulsion Technology, Vol. 1, Becher, P., Ed, Marcel Dekker, New York, 1983.Shoemaker, C.F., Nantz, J., Bonnans, S., and Noble, A.C., Rheological characterization of dairy

products, Food Technology, 46, 98, 1992.Siano, S.A., Applications development with radio-frequency dielectric spectroscopy: Emulsions, sus-

pensions and biomass, in Handbook on Ultrasonic and Dielectric Characterization Techniquesfor Suspended Particles, Hackley, V.A. and Texter, J., Eds., American Chemical Society,Westerville, OH, 1998.

Simatos, D. and Multon, J.L., Properties of Water in Foods in Relation to Quality and Stability,Martinus Nijhoff Publishers, Dordrecht, Netherlands, 1985.

Simic, M.G., Jovanovic, S.V., and Niki, E., Mechanisms of lipid oxidative processes and their inhibi-tion, in Lipid Oxidation in Food, St. Angelo, A.J., Ed., American Chemical Society, Washing-ton, D.C., 1992, chap. 2.


Simoneau, C., McCarthy, M.J., Kauten, R.J., and German, J.B., Crystallization dynamics in modelemulsions from magnetic resonance imaging, Journal of the American Oil Chemists Society,68, 481, 1991.

Simoneau, C., McCarthy, M.J., Reid, D.S., and German, J.B., Influence of triglyceride composition oncrystallization kinetics of model emulsions, Journal of Food Engineering, 19, 365, 1993.

Sjoblom, J., Fordedal, H., and Skodvin, T., Flocculation and coalescence in emulsions studied bydielectric spectroscopy, in Emulsion and Emulsion Stability, Sjoblom, J., Ed., Marcel Dekker,New York, 1996, chap. 8.

Skoda, W. and van den Tempel, M., Crystallization of emulsified triglycerides, Journal of ColloidScience, 18, 568, 1963.

Skodvin, T., Sjoblom, J., Saeten, J.O., Warnhim, T., and Gestblom, B., A time domain dielectricspectroscopy study of some model emulsions and liquid margarines, Colloids and Surfaces A,83, 75, 1994.

Skoog, D.A., West, D.M., and Holler, F.J., Analytical Chemistry: An Introduction, 6th ed., SaundersCollege Publishing, Philadelphia, 1994.

Sloan, A.E., Top 10 trends to watch and work on, Food Technology, 48(7), 89, 1994.Sloan, A.E., America’s appetite ’96: The top 10 trends to watch and work on, Food Technology, 50(7),

55, 1996.Smart, M.G., Fulcher, R.G., and Pechak, D.G., Recent developments in the microstructural character-

ization of foods, in Characterization of Foods: Emerging Techniques, Gaonkar, A.G., Ed.,Elsevier, Amsterdam, 1995, chap. 11.

Smith, A.L. and Mitchell, D.P., The centrifuge technique in the study of emulsion stability, in Theory andPractice of Emulsion Technology, Smith, A.L., Ed., Academic Press, London, 1976, chap. 4.

Smith, N.J. and Williams, P.A., Depletion flocculation of polystyrene lattices by water-soluble poly-mers, Journal of the Chemical Society, Faraday Transactions, 91, 1483, 1995.

Smolin, L.A. and Grosvenor, M.B., Nutrition: Science and Applications, Saunders College Publishing,Fort Worth, TX, 1994.

Soderberg, I., Hernqvist, L., and Bucheim, W., Milk fat crystallization in natural milk fat globules,Milchwissenschaft, 44, 403, 1989.

Soderman, O. and Bailnov, B., NMR self-diffusion studies of emulsions, in Emulsions and EmulsionStability, Sjoblom, J., Ed., Marcel Dekker, New York, 1996, chap. 8.

Soderman, O., Lonnqvist, I., and Bailnov, B., NMR self-diffusion studies of emulsion systems. Dropletsizes and microstructure of the continuous phase, in Emulsions — A Fundamental and PracticalApproach, Sjoblom, J., Ed., Kluwer, Dordrecht, Netherlands, 1992, 239.

Sonntag, N.O.V., Structure and composition of fats and oils, in Bailey’s Industrial Oil and Fat Prod-ucts, Vol. 1, Swern, D., Ed., John Wiley & Sons, New York, 1979a, chap. 1.

Sonntag, N.O.V., Reactions of fats and fatty acids, in Bailey’s Industrial Oil and Fat Products, Vol. 1,Swern, D., Ed., John Wiley & Sons, New York, 1979b, chap. 2.

Sonntag, N.O.V., Sources, utilization and classification of oils and fats, in Bailey’s Industrial Oil andFat Products, Vol. 1, Swern, D., Ed., John Wiley & Sons, New York, 1979c, chap. 5.

Sperry, P.R., A simple quantitative model for the volume restriction flocculation of latex by water-soluble polymers, Journal of Colloid Science, 87, 375, 1982.

Spicer, P.T. and Pratsinis, S.E., Shear-induced flocculation — The evolution of floc structure and theshape of the size distribution at steady-state, Water Research, 30, 1049, 1996.

Sridhar, S. and McDonald, M.J., Protein hydration investigations with high-frequency dielectric spec-troscopy, Journal of Physical Chemistry, 98, 6644, 1994.

Stang, M., Karbstein, H., and Schubert, H., Adsorption kinetics of emulsifiers at oil–water interfaces andtheir effect on mechanical emulsification, Chemical Engineering and Processing, 33, 307, 1994.

St. Angelo, A., brief introduction to food emulsions and emulsifiers, in Food Emulsifiers: Chemistry,Technology, Functional Properties and Applications, Charalambous, G. and Doxastakis, G.,Eds., Elsevier, Amsterdam, 1989.

St. Angelo, A.J., Lipid Oxidation in Food, American Chemical Society, Washington, D.C., 1992.Stauffer, C.E., Emulsifiers for the food industry, in Bailey’s Industrial Oil and Fat Products, Vol. 3,

5th ed., Hui, Y.H., Ed., John Wiley, New York, 1996, chap. 12.Stephan, A.M., Food Polysaccharides and Their Applications, Marcel Dekker, New York, 1995.



Stoll, S. and Buffle, J., Computer-simulation of bridging flocculation processes — The role of colloidto polymer concentration ratio on aggregation kinetics, Journal of Colloid and Interface Sci-ence, 180, 548, 1996.

Stone, H.A., Dynamics of drop deformation and breakup in viscous fluids, Annual Review of FluidMechanics, 26, 65, 1994.

Stothart, P.H., Developments in the application of small angle neutron scattering to food systems, inCharacterization of Foods: Emerging Techniques, Gaonkar, A.G., Ed., Elsevier, Amsterdam,1995, chap. 9.

Strawbridge, K.B., Ray, E., Hallett, F.R., Tosh, S.M., and Dalgleish, D.G., Measurement of the particlesize distribution in milk homogenized by a microfluidizer — Estimation of populations ofparticles with radii less than 100 nm, Journal of Colloid and Interface Science, 171, 392, 1995.

Suggett, A., Water–carbohydrate interactions, in Water Relationships of Foods, Duckworth, R.B., Ed,,Academic Press, London, 1975a, 23.

Suggett, A., Polysaccharides, in Water: A Comprehensive Treatise, Vol. 4, Franks, F., Ed., PlenumPress, New York, 1975b, chap. 6.

Swaisgood, H.E., Characteristics of milk, in Food Chemistry, 3rd ed., Fennema, G.R., Ed., MarcelDekker, New York, 1996, chap. 14.

Swift, C.E., Lockett, C., and Fryer, P.J., Comminuted meat emulsions — The capacity of meat foremulsifying fat, Food Technology, 15, 469, 1961.

Tadros, T.F., Fundamental principles of emulsion rheology and their applications, Colloids and Sur-faces, 91, 30, 1994.

Tadros, Th.F., Correlation of viscoelastic properties of stable and flocculated suspensions with theirinterparticle interactions, Advances in Colloid and Interface Science, 68, 97, 1996.

Tadros, T.F. and Vincent, B., Liquid/liquid interfaces, in Encyclopedia of Emulsion Technology, Vol.1, Becher, P., Ed., Marcel Dekker, New York, 1983.

Taisne, L., Walstra, P., and Cabane, B., Transfer of oil between emulsion droplets, Journal of Colloidand Interface Science, 184, 378, 1996.

Tan, C.-T., Beverage emulsions, in Food Emulsions, 2nd ed., Larsson, K. and Friberg, S.E., Eds.,Marcel Dekker, New York, 1990, chap. 10.

Tanford, C., The Hydrophobic Effect, Wiley, New York, 1980.Taylor, P., Ostwald ripening in emulsions, Colloids and Surfaces A, 99, 175, 1995.Taylor, A.J. and Linforth, R.S.T., Flavour release in the mouth, Trends in Food Science and Technol-

ogy, 7, 444, 1996.Texter, J., Beverly, T., Templar, S.R., and Matsubara, T., Partitioning of para-phenylenediamines in oil-

in-water emulsions, Journal of Colloid and Interface Science, 120, 389, 1987.Thomson, D.M.H., The meaning of flavor, in Developments in Food Flavours, Birch, G.G. and Lindley,

M.G., Eds., Elsevier Applied Science, London, 1986, 1.Timms, R.E., Crystallization of fats, Chemistry and Industry, p. 342, May 1991.Timms, R.E., Crystallization of fats, in Developments in Oils and Fats, Hamilton, R.J., Ed., Blackie

Academic and Professional, London, 1995, chap. 8.Tinoco, I., Sauer, K., and Wang, J.C., Physical Chemistry: Principles and Applications in Biological

Sciences, 2nd ed., Prentice-Hall, Englewood Cliffs, NJ, 1985.Tolstoguzov, V., Structure–property relationships in foods, in Macromolecular Interactions in Food

Technology, Parris, N., Kato, A., Creamer, L.K., and Pearce, J., Eds., American ChemicalSociety, Washington, D.C., 1996, chap. 1.

Tornberg, E. and Hermannson, A.M., Functional characteristics of protein stabilized emulsions. Effectof processing, Journal of Food Science, 42, 468, 1977.

Tory, E.M., Sedimentation of Small Particles in a Viscous Liquid, Computational Mechanics Publica-tions, Southampton, U.K., 1996.

Trainer, M.N., Freud, P.J., and Leonardo, E.M., High-concentration submicron particle size distributionby dynamic light scattering, American Laboratory, p. 34, July 1992.

Tung, M.A. and Jones, L.J., Microstructure of mayonnaise and salad dressing, in Studies of FoodMicrostructure, Holcomb, D.N. and Kalab, M., Eds., Scanning Electron Microscopy, AMFO’Hare, IL, 1981, 231.


Tung, M.A. and Paulson, A.T., Rheological concepts for probing ingredient interactions in foodsystems, in Ingredient Interactions: Effects on Food Quality, Gaonkar, A., Ed., Marcel Dekker,New York, 1995.

Tunon, I., Silla, E., and Pascual-Ahuir, J.L., Molecular surface area and hydrophobic effect, ProteinEngineering, 5, 715, 1992.

Turnbull, D. and Cormia, D., Kinetics of crystal nucleation in some normal alkane liquids, Journal ofChemical Physics, 34, 820, 1961.

Uriev, N.B., Structure, rheology and stability of concentrated disperse systems under dynamic condi-tions, Colloids and Surfaces A, 87, 1, 1994.

Vaessen, G.E.J. and Stein, H.N., The applicability of catastrophe theory to emulsion phase inversion,Journal of Colloid and Interface Science, 176, 378, 1995.

van Boekel, M.A.J.S., Estimation of solid liquid ratios in bulk fats and emulsions by pulsed NMR,Journal of the American Oil Chemists Society, 58, 768, 1981.

van Boekel, M.A.J.S. and Walstra, P., Stability of oil-in-water emulsions with crystals in the dispersephase, Colloids and Surfaces, 3, 109, 1981.

van de Hulst, H.C., Light Scattering by Small Particles, John Wiley & Sons, New York, 1957.van den Enden, J.C., Waddington, D., van Aalst, H., van Kralingen, C.G., and Packer, K.J., Rapid

determination of water droplet size distributions by PFG-NMR, Journal of Colloid and Inter-face Science, 140, 105, 1990.

van Gunsteren, W.F., The role of computer simulation techniques in protein engineering, ProteinEngineering, 2, 5, 1988.

van Holde, K.E., Effects of amino acid composition and microenvironment on protein structure, in FoodProteins, Whitaker, J.R. and Tannenbaum, S.R., Eds., AVI Publishing, Westport, CT, 1977,chap. 1.

van Vliet, T., Mechanical properties of concentrated food gels, in Food Macromolecules and Colloids,Dickinson, E. and Lorient, D., Eds., Royal Society of Chemistry, Cambridge, 1995, 447.

van Vliet, T. and Walstra, P., Weak particle networks, in Food Colloids, Bee, R.D., Richmond, P., andMingins, J., Eds., Royal Society of Chemistry, Cambridge, 1989, 206.

Vaughan, J.G., Food emulsions, in Food Science and Technology. A Series of Monographs. FoodMicroscopy, Vaughan, J.G., Ed., Academic Press, London, 1979.

Vodovotz, Y., Vittadini, E., Coupland, J., McClements, D.J., and Chinachoti, P., Bridging the gap: Useof confocal microscopy in food research, Food Technology, 50, 74, 1996.

Vold, M.J., The effect of adsorption on the van der Waals interaction of spherical colloidal particles,Journal of Colloid Science, 16, 1, 1961.

Waddington, D., Some applications of wide line NMR in the oils and fats industry, in Fats and Oils:Chemistry and Technology, Hamilton, R.J. and Bhati, A., Eds., Applied Science Publishers,London, 1980, chap. 2.

Walstra, P., On the crystallization habit in fat globules, Netherlands Milk and Dairy Journal, 21, 166, 1967.Walstra, P., Estimating globule-size distribution of oil-in-water emulsions by spectroturbidimetry,

Journal of Colloid and Interface Science, 27, 493, 1968.Walstra, P., Formation of emulsions, in Encyclopedia of Emulsion Technology, Vol. 1, Becher, P., Ed.,

Marcel Dekker, New York, 1983, chap. 2.Walstra, P., Fat crystallization, in Food Structure and Behaviour, Blanshard, J.M.V. and Lillford, P.,

Eds., Academic Press, London, 1987, chap. 5.Walstra, P., Introduction to aggregation phenomenon in food colloids, in Food Colloids and Polymers:

Stability and Mechanical Properties, Dickinson, E. and Walstra, P., Eds., Royal Society ofChemistry, Cambridge, 1993a, 3.

Walstra, P., Principles of emulsion formation, Chemical Engineering Science, 48, 333, 1993b.Walstra, P., Emulsion stability, in Encyclopedia of Emulsion Technology, Vol. 4, Becher, P., Ed.,

Marcel Dekker, New York, 1996a, chap. 1.Walstra, P., Disperse systems: Basic considerations, in Food Chemistry, 3rd ed., Fennema, O.R., Ed.,

Marcel Dekker, New York, 1996b, chap. 3.Walstra, P. and van Beresteyn, E.C.H., Crystallization of milk fat in the emulsified state, Netherlands

Milk and Dairy Journal, 29, 35, 1975.



Wedzicha, B.L., Distribution of low-molecular weight food additives in dispersed systems, in Advancesin Food Emulsions, Dickinson, E. and Stainsby, G., Eds., Elsevier, London, 1988, chap. 10.

Wedzicha, B.L., Zeb, A., and Ahmed, S., Reactivity of food preservatives in dispersed systems, in FoodPolymers, Gels and Colloids, Dickinson, E., Ed., Royal Society of Chemistry, Cambridge,1991, 180.

Wehling, R.L., Infrared spectroscopy, in Introduction to the Chemical Analysis of Foods, Nielsen, S.S.,Ed., Bartlett and Jones, Boston, 1994, chap. 24.

Wei, Y.-Z., Kumbharkhane, A.C., Sadeghi, M., Sage, J.T., Tiam, W.D., Champion, P.M., Reiner, E.S.,and Radke, C.J., Double layer interactions between charge-regulated colloidal systems: Pairpotentials for spherical particles bearing ionogenic surface groups, Advances in Colloid andInterface Science, 47, 59, 1993.

Wei, Y.-Z., Kumbharkhane, A.C., Sadeghi, M., Sage, J.T., Tiam, W.D., Champion, P.M., Radke, C.J.,and McDonald, M.J., Protein hydration investigations with high-frequency dielectric spectros-copy, Journal of Physical Chemistry, 98, 6644, 1994.

Weiss, T.J., Food Oils and Their Uses, 2nd ed., AVI Publishing, Westport, CT, 1983.Weiss, J., Coupland, J.N., and McClements, D.J., Solubilization of hydrocarbon droplets suspended in

a non-ionic surfactant solution, Journal of Physical Chemistry, 100, 1066, 1996.Whitaker, J.R., Denaturation and renaturation of proteins, in Food Proteins, Whitaker, J.R. and

Tannenbaum, S.R., Eds., AVI Publishing, Westport, CT, 1977, chap. 2.Whitaker, J.R. and Tannenbaum, S.R., Eds., Food Proteins, AVI Publishing, Westport, CT, 1977.Whorlow, R.W., Rheological Techniques, 2nd ed., Ellis Horwood, New York, 1992.Williams, A.A., Modern sensory analysis and the flavour industry, in Developments in Food Flavours,

Birch, G.G. and Lindley, M.G., Eds., Elsevier Applied Science, London, 1986, 1.Williams, A., Janssen, J.J.M., and Prins, A., Behaviour of droplets in simple shear flow in the presence

of a protein emulsifier, Colloids and Surfaces A, 125, 189, 1997.Wilson, R.H., Recent developments in infrared spectroscopy and microscopy, in New Physicochemical

Techniques for the Characterization of Complex Food Systems, Dickinson, E., Ed., BlackieAcademic and Professional, London, 1995, chap. 8.

Wunderlich, B., Thermal Analysis, Academic Press, New York, 1990.Xang, H.J. and Xu, G.D., The effect of particle refractive index on size measurement, Powder Tech-

nology, 70, 189, 1992.Xiong, Y.L. and Kinsella, J.E., Influence of fat globule membrane composition and fat type on the

rheological properties of milk based composition gels. 1. Methodology, Milchwissenschaft, 46,150, 1991.

Xiong, Y.L., Aguilera, J.M., and Kinsella, J.E., Emulsified milkfat effects on rheology of acid-inducedmilk gels, Journal of Food Science, 56, 920, 1991.

Yaminsky, V.V., Ninham, B.W., Christenson, H.K., and Pashley, R.M., Adsorption forces betweenhydrophobic monolayers, Langmuir, 12, 1936, 1996a.

Yaminsky, V.V., Jones, C., Yaminsky, F., and Ninham, B.W., Onset of hydrophobic attraction at lowsurfactant concentrations, Langmuir, 12, 3531, 1996b.

Yost, R.A. and Kinsella, J.E., Properties of acidic whey protein gels containing emulsified butterfat,Journal of Food Science, 57, 158, 1993.

Young, S.L., Sarda, X., and Rosenberg, M., Microencapsulation properties of whey proteins. 1.Microencapsulation of anhydrous milk fat, Journal of Dairy Science, 76, 2868, 1993a.

Young, S.L., Sarda, X., and Rosenberg, M., Microencapsulation properties of whey proteins. 2. Com-bination of whey proteins with carbohydrates, Journal of Dairy Science, 76, 2878, 1993b.

Zhang, X. and Davis, H.D., The rate of collisions due to Brownian or gravitational motion on smalldroplets, Journal of Fluid Mechanics, 230, 479, 1991.

Zhang, H.J. and Xu, G.D., The effect of particle refractive index on size measurement, PowderTechnology, 70, 189, 1992.

Ziegler, G.R. and Foegedding, E.A., The gelation of proteins, in Advances in Food and NutritionResearch, Vol. 34, Kinsella, J.E., Ed., Academic Press, San Diego, 1990, 203.

Zielinski, R.J., Synthesis and composition of food-grade emulsifiers, in Food Emulsifiers and TheirApplications, Hasenhuettl, G.L. and Hartel, R.W., Eds., Chapman and Hall, New York, 1997,chap. 2.


Documents

Food Emulsions